13.2 Qualitative & Quantitative Risk Analysis

13.2 Qualitative & Quantitative Risk Analysis

Quick Summary: Once project risks are identified in the Risk Register, they must be analyzed to guide decision-making and allocate capital. Risk analysis divides into two complementary disciplines: Qualitative Risk Analysis, which rapidly evaluates probability and impact using ordinal scales (1–5) and a P-I Matrix to rank risks into Red, Amber, and Green zones; and Quantitative Risk Analysis, which statistically models overall project cost and schedule uncertainty. Quantitative methods include Expected Monetary Value (EMV), Decision Tree modeling, Tornado sensitivity analysis, and Monte Carlo simulation. Monte Carlo simulation generates an S-curve (CDF) that establishes probabilistic confidence targets (e.g., P50, P80) and dictates contingency reserves.

---

1. Qualitative vs. Quantitative Analysis: Structural Comparison

A central competency tested on the AACE CCT exam is distinguishing between qualitative and quantitative risk assessments across operational scope, input data, and analytical outputs.

+-----------------------------------------------------------------------------------+
|              QUALITATIVE VS. QUANTITATIVE RISK ANALYSIS COMPARISON                |
|                                                                                   |
|  FEATURE             QUALITATIVE ANALYSIS            QUANTITATIVE ANALYSIS        |
|  -------------------------------------------------------------------------------  |
|  Primary Goal        Fast prioritization & ranking   Probabilistic cost & schedule|
|                      of individual risks             forecasting for overall asset|
|                                                                                   |
|  Analytical Method   Ordinal scoring (1-5, Low/Med)  Stochastic modeling, EMV,    |
|                      Probability-Impact (P-I) Matrix Monte Carlo, Decision Trees  |
|                                                                                   |
|  Input Data          Subjective expert judgment      Empirical historical data,   |
|                      & workshop consensus            3-point ranges, distributions|
|                                                                                   |
|  Primary Output      Prioritized Risk Register       Cumulative S-Curves (CDF),   |
|                      (Red / Amber / Green zones)     P50/P80 values, dollar reserves|
|                                                                                   |
|  Execution Cadence   Performed on ALL identified     Performed on major/complex   |
|                      risks; repeated frequently      projects; requires more effort|
+-----------------------------------------------------------------------------------+

---

2. Qualitative Risk Assessment & The P-I Scoring Matrix

Qualitative risk analysis evaluates two fundamental dimensions for each identified risk event:

1. Probability ($P$): The likelihood that the uncertain event will occur (expressed as a probability percentage or ordinal rating).
2. Impact ($I$): The severity of the consequence if the event occurs, evaluated across cost, schedule, quality, and safety.

Mathematical Risk Score Formulation

In qualitative evaluation, the composite Risk Score is computed as:

$\text{Risk Score} = \text{Probability Rating} \times \text{Impact Rating}$

Ordinal scales typically range from 1 to 5 (or Very Low, Low, Moderate, High, Very High). To eliminate subjective bias during risk workshops, the Risk Management Plan must establish objective, calibrated definitions for each rating level.

Objective Ordinal Calibration Scale (Illustrative Standard):

The 5x5 Probability-Impact (P-I) Heat Map

The intersection of probability and impact defines the risk position on the P-I Matrix, categorizing risks into three distinct management action zones:

+-----------------------------------------------------------------------------------+
|                         5x5 PROBABILITY-IMPACT (P-I) MATRIX                       |
|                                                                                   |
|  PROBABILITY                                                                      |
|  Very High (5) |  [ 5 ]    [ 10 ]    [ 15 ]    [ 20 ]    [ 25 ]                   |
|  High (4)      |  [ 4 ]    [  8 ]    [ 12 ]    [ 16 ]    [ 20 ]                   |
|  Moderate (3)  |  [ 3 ]    [  6 ]    [  9 ]    [ 12 ]    [ 15 ]                   |
|  Low (2)       |  [ 2 ]    [  4 ]    [  6 ]    [  8 ]    [ 10 ]                   |
|  Very Low (1)  |  [ 1 ]    [  2 ]    [  3 ]    [  4 ]    [  5 ]                   |
|                +-----------------------------------------------                   |
|                    (1)       (2)       (3)       (4)       (5)                    |
|                  V.Low       Low     Moderate   High      V.High                  |
|                                   IMPACT                                          |
|                                                                                   |
|  ZONE CLASSIFICATION & ACTION THRESHOLDS:                                         |
|  * RED ZONE (High Risk: Scores 15 - 25): Critical priority. Active mitigation,   |
|    mandatory executive oversight, and targeted contingency allocation required.   |
|  * AMBER ZONE (Moderate Risk: Scores 6 - 14): Moderate priority. Proactive       |
|    monitoring, periodic reassessment, and secondary response plans established.   |
|  * GREEN ZONE (Low Risk: Scores 1 - 5): Low priority. Placed on a passive        |
|    Risk Watch List; accepted without dedicated capital expenditure.               |
+-----------------------------------------------------------------------------------+

Multi-Dimensional Qualitative Assessment

Beyond Probability and Impact, advanced TCM practice incorporates three qualitative evaluation parameters:

• Risk Urgency / Proximity: How soon the risk event could occur. A moderate-impact risk occurring next week demands higher immediate attention than a high-impact risk three years away.
• Risk Detectability: How easily the trigger condition can be recognized before the risk materializes. Low detectability increases vulnerability.
• Propinquity: The degree to which the risk directly impacts core strategic objectives of executive stakeholders.

---

3. Quantitative Risk Analysis: Expected Monetary Value (EMV) & Decision Trees

Quantitative Risk Analysis uses mathematical and probabilistic models to quantify project uncertainty in dollars or calendar days.

Expected Monetary Value (EMV)

Expected Monetary Value (EMV) is a statistical calculation that determines the probability-weighted average outcome of an uncertain event:

$\text{EMV} = \text{Probability} (P) \times \text{Monetary Impact} (I)$

$\text{Total Project EMV} = \sum_{i=1}^{n} (P_i \times I_i)$

• For Threats: Monetary impact is negative (cost increase or cash outflow), yielding a negative EMV.
• For Opportunities: Monetary impact is positive (cost savings or revenue gain), yielding a positive EMV.

Decision Tree Modeling Under Risk

A Decision Tree is a graphical diagram that models a sequence of decisions and probabilistic chance outcomes, allowing cost engineers to determine the financially optimal strategy.

+-----------------------------------------------------------------------------------+
|                         DECISION TREE NOTATION & SYMBOLS                          |
|                                                                                   |
|  [ [] ] DECISION NODE (Square): Represents a point where the project team has    |
|         direct management control to choose among competing alternatives.         |
|                                                                                   |
|  [ (O) ] CHANCE NODE (Circle): Represents an uncertain event beyond direct       |
|          control. Each branch has an assigned probability (Sum of P = 1.0) and    |
|          a conditional financial payoff.                                          |
|                                                                                   |
|  [ |> ] END NODE (Triangle): Represents the final terminal financial outcome      |
|         of a specific path through the decision tree.                             |
+-----------------------------------------------------------------------------------+

Comprehensive Worked Mathematical Example: Foundation Piling Selection

An EPC contractor must select a foundation installation technique for a subsea marine terminal. Soil borings show variable geotechnical conditions.

• Alternative A: Driven Precast Concrete Piles

  • Fixed Capital Mobilization & Installation Cost: $1,200,000
  • Geotechnical Uncertainty (Chance Node A):
    • 75% Probability ($P_1 = 0.75$): Normal soils encountered; pile driving completes without obstruction ($I_1 = $0$ additional cost).
    • 25% Probability ($P_2 = 0.25$): Subsurface boulder refusal encountered; requires heavy drilling rigs and remedial coring at an extra cost of $1,600,000 ($I_2 = -$1,600,000$).

• Alternative B: Bored Cast-In-Place Shafts (Drilled Shafts)

  • Fixed Capital Mobilization & Installation Cost: $1,500,000
  • Geotechnical Uncertainty (Chance Node B):
    • 85% Probability ($P_1 = 0.85$): Stable excavation sidewalls; shafts poured cleanly without casing failure ($I_1 = $0$ additional cost).
    • 15% Probability ($P_2 = 0.15$): Caving soils encountered; requires pressurized slurry stabilization and steel liner casings at an extra cost of $800,000 ($I_2 = -$800,000$).

Step-by-Step Decision Tree Mathematical Derivation:

Step 1: Calculate the Expected Monetary Value (EMV) at Chance Node A (Driven Piles) $\text{EMV}_{\text{Chance A}} = (0.75 \times \$0) + (0.25 \times \$1,600,000) = \$0 + \$400,000 = \$400,000$ $\text{Total Expected Cost (Alternative A)} = \text{Initial Cost} + \text{EMV}_{\text{Chance A}} = \$1,200,000 + \$400,000 = \$1,600,000$

Step 2: Calculate the Expected Monetary Value (EMV) at Chance Node B (Drilled Shafts) $\text{EMV}_{\text{Chance B}} = (0.85 \times \$0) + (0.15 \times \$800,000) = \$0 + \$120,000 = \$120,000$ $\text{Total Expected Cost (Alternative B)} = \text{Initial Cost} + \text{EMV}_{\text{Chance B}} = \$1,500,000 + \$120,000 = \$1,620,000$

Step 3: Economic Decision Synthesis

• Alternative A (Driven Piles) Total Expected Cost = $1,600,000
• Alternative B (Drilled Shafts) Total Expected Cost = $1,620,000
• Decision: Based strictly on Expected Monetary Value (risk-neutral posture), the project manager selects Alternative A (Driven Piles) because it delivers a net expected financial savings of $20,000 ($$1,620,000 - $1,600,000$).

[!TIP] The Risk Appetite Nuance: If the contractor is extremely risk-averse, Alternative B might still be chosen because its worst-case scenario is $2,300,000 ($1.5M + $800k), whereas Alternative A has a worst-case scenario of $2,800,000 ($1.2M + $1.6M). EMV assumes a risk-neutral decision-maker.

---

4. Sensitivity Analysis & The Tornado Diagram

Sensitivity Analysis evaluates the extent to which project uncertainties impact the project outcome by varying one input parameter across its credible range while holding all other parameters constant at their baseline values.

The Tornado Diagram Mechanics

A Tornado Diagram is a specialized horizontal bar chart used to display the results of sensitivity analysis:

• Each horizontal bar represents an individual risk variable or cost account.
• The length of the bar represents the range (swing) between the pessimistic outcome and the optimistic outcome.
• The vertical centerline represents the deterministic base estimate.
• Bars are arranged in descending order of range width: the variable causing the largest variance is placed at the top, and subsequent variables taper downward, forming an inverted funnel shape resembling a tornado.

+-----------------------------------------------------------------------------------+
|                    TYPICAL PROJECT TORNADO SENSITIVITY DIAGRAM                    |
|                                                                                   |
|  Base Estimate = $10,000,000                                                      |
|                                                                                   |
|  Risk Variable                    Low Impact               High Impact     Spread |
|  -------------------------------------------------------------------------------  |
|  Piping Craft Labor Productivity    [-$600k] ================== [+$1,400k]  $2.0M |
|  Structural Steel Market Price        [-$400k] ============= [+$800k]       $1.2M |
|  Subsurface Bedrock Remediation        [-$100k] ======== [+$700k]           $800k |
|  Equipment Ocean Freight Shipping       [-$150k] ===== [+$350k]             $500k |
|  Electrical Cable Tray Rework             [-$50k] == [+$150k]               $200k |
|                                                                                   |
|                                          |                                        |
|                                    Base ($10.0M)                                  |
+-----------------------------------------------------------------------------------+

Practical Utility of the Tornado Diagram:

1. Focuses Resources on the "Vital Few": Demonstrates that managing labor productivity and steel commodity prices provides far greater risk reduction than managing electrical tray rework.
2. Eliminates False Precision: Exposes which deterministic assumptions carry the greatest downside exposure.

---

5. Monte Carlo Simulation Principles & S-Curve Percentiles

Monte Carlo simulation is a computerized mathematical technique that quantifies risk and uncertainty by calculating project outcomes across thousands of simulated execution trials (typically 1,000 to 10,000 iterations).

Input Probability Distributions

Instead of entering fixed single-point values (deterministic estimating), cost engineers define probability distributions for uncertain cost line items and activity durations:

+-----------------------------------------------------------------------------------+
|                     COMMON PROBABILITY DISTRIBUTION TYPES                         |
|                                                                                   |
|  1. TRIANGULAR DISTRIBUTION:           2. BETA / PERT DISTRIBUTION:               |
|     Defined by Minimum (a), Most          Smooth bell-shaped curve with tails.    |
|     Likely (m), and Maximum (b).          Heavily weights the most likely value.  |
|          /\                                    _---_                              |
|         /  \                                 /       \                            |
|       /      \                             /           \                          |
|     +----------+                         +---------------+                        |
|     a    m     b                         a       m       b                        |
|     (Widely used when historical         (Standard in CPM scheduling &            |
|     empirical data is scarce)            three-point PERT calculations)           |
|                                                                                   |
|  3. UNIFORM DISTRIBUTION:              4. LOGNORMAL DISTRIBUTION:                 |
|     Equal probability across range.       Positively skewed with a long right tail.|
|     +----------+                          Bounded at zero (costs cannot be negative)|
|     |          |                             /\_                                  |
|     |          |                            /   ---_                              |
|     +----------+                          /         ---_                          |
|     a          b                         +------------------+                     |
+-----------------------------------------------------------------------------------+

Sampling Techniques: Random vs. Latin Hypercube

• Simple Random Sampling (Standard Monte Carlo): Pure random number generation. Requires 10,000+ iterations because samples can cluster, leaving portions of the distribution unsampled.
• Latin Hypercube Sampling (LHS): A stratified sampling technique that stratifies the input distribution into intervals of equal probability and ensures each interval is sampled evenly. LHS achieves equivalent statistical stability with far fewer iterations (e.g., 1,000 to 2,000 runs).

Simulation Outputs: PDF vs. CDF (The S-Curve)

The simulation aggregates thousands of iterations into two distinct statistical curves:

1. Probability Density Function (PDF): A relative frequency histogram showing the probability of achieving any specific dollar value or completion date.
2. Cumulative Distribution Function (CDF / S-Curve): A sigmoidal S-shaped curve showing the cumulative probability of completing the project at or below a given cost or date.

+-----------------------------------------------------------------------------------+
|                    CUMULATIVE DISTRIBUTION S-CURVE & PERCENTILES                  |
|                                                                                   |
|  Cumulative                                                                       |
|  Probability                                                                      |
|  100% |                                                    ..-==+ (P100)          |
|       |                                              .---''                       |
|   80% | - - - - - - - - - - - - - - - - - - - -+---'' (P80: Risk-Adjusted Baseline)|
|       |                                    _.-'                                   |
|   50% | - - - - - - - - - - - - - - - - +-' (P50: Median Outcome)                 |
|       |                             _.-'                                          |
|   20% |                         _.-'                                              |
|   10% | - - - - - - - - - - -+-' (P10: Optimistic / Aggressive Target)             |
|    0% +----------------------+---------------------------------------             |
|                           Base      P50          P80                              |
|                         Estimate   ($10.8M)    ($11.6M)                           |
|                         ($10.0M)                                                  |
|                                                                                   |
|  CONTINGENCY SIZING FORMULA:                                                      |
|  * Base Estimate (Deterministic): $10,000,000                                     |
|  * P80 Budget (80% Confidence):   $11,600,000                                     |
|  * Required Contingency Reserve:  $11,600,000 - $10,000,000 = $1,600,000 (16.0%)  |
+-----------------------------------------------------------------------------------+

High-Yield Percentile Interpretation:

• P10 Value: Optimistic / aggressive target. There is only a 10% probability that final cost will be at or below this number (a 90% risk of cost overrun).
• P50 Value (Median): Represents the 50th percentile. There is an equal 50% probability that the project will underrun or overrun this cost.
• P80 Value: Conservative, risk-adjusted budget target commonly mandated by capital asset owners and government agencies. Represents an 80% probability of completing at or below this value.
• Contingency Determination Formula:

$\text{Contingency Reserve} = P_{80} \text{ Cost} - \text{Deterministic Base Estimate}$

Limitations & Data Requirements of Quantitative Models

While Monte Carlo simulation provides mathematical rigor, it carries critical limitations:

1. Garbage In, Garbage Out (GIGO): The output is only as valid as the underlying ranges and distribution assumptions.
2. Correlation Neglect: In reality, cost items correlate (e.g., if concrete volume increases, concrete finishing labor hours and rebar tonnage also increase). If the cost engineer fails to input correlation coefficients, the simulation will erroneously assume variances cancel each other out, severely understating total project risk!

---

6. Exam Watch: High-Yield Traps & Rules of Thumb

[!WARNING] The Ordinal Score Addition Fallacy: In qualitative risk analysis, never add ordinal risk scores together across risks (e.g., adding a score of 20 for Risk A and 15 for Risk B to get 35). Ordinal ratings are relative rankings, not absolute scalar quantities! They cannot be manipulated through algebraic summation.

[!CAUTION] The Deterministic Base Estimate Is NOT P50: In construction estimating, the deterministic base estimate almost never represents the P50 median! Because real-world cost distributions are right-skewed (costs can overrun by 100%, but cannot underrun by 100%), the deterministic estimate typically sits around the P15 to P30 percentile. It is inherently biased toward underestimation without contingency.

[!TIP] Rolling Back Decision Trees: When solving decision tree questions, always solve backwards from right to left (rolling back the tree). Calculate the EMV of chance nodes at the far right first, and then work back to the decision nodes on the left.