Forecasting, Change Control and Corrective Decisions

Introduction and Context

Project cost management does not end when the approved budget is established. During execution, actual costs, completed work, market conditions, productivity, scope interpretation and risk exposure continually change. Managers must therefore answer four connected questions:

1. What has the project achieved relative to the plan?
2. What is the likely final cost and completion date?
3. Which changes should be approved, rejected or deferred?
4. What corrective action can protect value, cash flow and strategic objectives?

Forecasting converts current performance information into an estimate of future outcomes. Change control determines whether the approved cost, scope and schedule baselines should be altered. Corrective decisions address unacceptable deviations, while cost recovery seeks to preserve or restore the economic position of the project. These activities are interdependent: an unreliable forecast weakens change evaluation, and uncontrolled change makes subsequent forecasts meaningless.

Learning Objectives

By the end of this chapter, the learner should be able to:

• calculate and interpret estimate at completion, estimate to complete and variance at completion;
• select appropriate forecasting methods according to the cause and persistence of variance;
• use trends, scenarios and sensitivity analysis to support managerial decisions;
• distinguish corrective action, preventive action, defect repair and rebaselining;
• evaluate change requests through integrated change control;
• design governance arrangements for cost recovery and ethical cost reporting; and
• assess the limitations of quantitative forecasts in uncertain project environments.

1. The Forecasting Architecture

A credible forecast is a controlled statement about a future outcome based on available evidence. It is not a target, a revised budget, or an optimistic opinion. Forecasting normally combines:

• the approved performance measurement baseline;
• actual cost incurred;
• physical progress and accepted deliverables;
• commitments and open purchase orders;
• remaining work and resource productivity;
• approved changes and unresolved claims;
• risk and opportunity exposure; and
• assumptions about prices, exchange rates, productivity and schedule.

A useful forecast distinguishes three financial quantities:

Forecasts should be time-phased and responsibility-based. A single project-wide number can conceal a serious problem in a work package, supplier contract or cost category. The cost engineer or project controller should maintain a forecast by work breakdown structure element, cost type, control account and reporting period, then reconcile the detail to the total project forecast.

2. Earned Value Measures for Forecasting

Earned value management (EVM) integrates scope, schedule and cost information. Its fundamental measures are:

The cost variance is:

The cost performance index is:

When CPI > 1, cost performance is favourable; when CPI < 1, the project is obtaining less budgeted work per unit of actual cost. The schedule variance and schedule performance index are:

In EVM, schedule variance is expressed in currency, not calendar time. It indicates whether the value of completed work differs from the value planned, but it does not directly state how many days late the project is. Schedule forecasting should therefore supplement EVM with network analysis, milestone trends and critical-path information.

These indicators assume that the baseline is logically sound, work is objectively measurable and actual costs are recorded in the same accounting period as performance. They are weak when scope is ambiguous, progress is estimated subjectively or costs are posted with long delays.

3. Estimate at Completion and Related Forecasts

The estimate to complete (ETC) is the expected additional cost required to finish all remaining work. The fundamental identity is:

The variance at completion (VAC) is:

A negative VAC indicates that the forecast exceeds the approved budget. These identities are general; the quality of EAC depends on the method used to estimate ETC.

3.1 Independent bottom-up ETC

The most defensible approach is often to estimate each remaining work package and add known commitments, expected escalation, unresolved liabilities and quantified risk exposure:

Then:

This method is appropriate when the project has experienced a material change, current performance is not representative of future work, or remaining tasks are sufficiently understood. It is resource-intensive and can be biased if managers deliberately understate remaining work.

3.2 EAC based on current cost efficiency

If current cost efficiency is expected to continue:

Here, BAC − EV is the budgeted value of remaining work. This method assumes that the observed cost performance will continue and that the remaining scope is comparable to completed scope.

3.3 EAC based on both cost and schedule efficiency

If remaining cost will be affected by both cost inefficiency and schedule inefficiency, a common formulation is:

This is not automatically superior to the CPI-only method. It assumes that schedule performance has a causal and continuing effect on cost. For example, a delay may create overtime and escalation, but an administrative delay with no cost consequence should not be mechanically inserted into the equation.

3.4 EAC assuming an atypical variance

If past variance is judged exceptional and the original estimate for remaining work remains valid:

Since AC + (BAC − EV) equals BAC − CV, this approach effectively treats the current cost variance as non-recurring. It is appropriate only when there is documented evidence, such as a one-time equipment failure that has been repaired or a non-repeatable accounting adjustment.

3.5 Variance at completion and required performance

The cost performance required to meet the original BAC is the to-complete performance index (TCPI):

If the approved forecast EAC is the realistic target instead, then:

The numerator is the budgeted value of remaining work. The denominator is the remaining allowable cost. TCPI is dimensionless and assumes that EV, AC and BAC are measured using the same currency and accounting basis. A very high TCPI may reveal that the original budget is no longer credible rather than motivate unrealistic pressure on the team.

Worked Example 1: EVM Forecasts

An infrastructure project has BAC = $10.0 million. At the status date, PV = $6.0 million, EV = $5.0 million, and AC = $5.8 million.

First:

The project is cost inefficient: approximately $0.862 of budgeted value is being produced for each $1.00 spent.

Using CPI-only continuation:

Therefore:

The forecast overrun is approximately $1.60 million. The required performance against the original budget is:

The remaining work would have to be delivered at a CPI of approximately 1.19, materially better than the current 0.862. Management should investigate the cause before treating the original BAC as achievable.

4. Trend Projection and Scenario Forecasting

A forecast should not rely on one indicator. Trend analysis examines performance over successive reporting periods. Useful approaches include:

• rolling CPI and SPI charts;
• cumulative cost curves compared with baseline;
• milestone slippage and recovery trends;
• productivity rates, such as units per labour-hour;
• committed-cost burn-down; and
• supplier price and claim trends.

A simple linear trend may be represented as:

where y_t is the forecast measure at period t, a is the intercept and b is the estimated change per period. Linear projection is suitable only when historical observations are reasonably stable and the driver is expected to remain linear. Construction productivity, commodity prices and risk occurrence often do not satisfy this condition.

Scenario forecasting is more appropriate when uncertainty is material. A practical set is:

• optimistic: favourable productivity, timely approvals and no major residual risks;
• most likely: current evidence and management’s central assumptions; and
• pessimistic: adverse productivity, escalation, delay and selected risk events.

An expected-value forecast can be calculated when scenario probabilities are credible:

where p_i is the probability of scenario i, and the probabilities sum to 1. Expected value is useful for portfolio planning, but it is not necessarily a likely project outcome. A project manager should also report the range, key assumptions and downside exposure.

Sensitivity analysis changes one assumption at a time, such as labour productivity or exchange rate. Scenario analysis changes a coherent set of assumptions. Monte Carlo simulation extends this by assigning probability distributions to uncertain variables and repeatedly calculating possible completion costs. Simulation is valuable for contingency and confidence-level decisions, but it cannot correct poor input data or false correlations.

5. Integrated Change Control

A change is any modification to approved scope, deliverables, specifications, design, schedule, cost, procurement arrangement or acceptance criteria. Integrated change control evaluates the total effect rather than considering cost in isolation.

A formal change process normally includes:

1. Initiation: record the requested change, origin, reason and urgency.
2. Screening: determine whether it is a valid change, clarification, defect correction or already-authorized work.
3. Impact analysis: estimate effects on scope, cost, schedule, quality, resources, risk, contracts, operations, sustainability and benefits.
4. Options analysis: compare alternatives, including doing nothing.
5. Authority review: obtain approval from the delegated change authority or change control board.
6. Decision communication: document approval, rejection, deferral, conditions and responsibilities.
7. Baseline update: revise only the affected controlled documents and retain the original version.
8. Implementation and verification: authorize work, update forecasts and confirm that the change delivered its intended result.

The change request should contain a unique identifier, description, business rationale, urgency, affected requirements, cost estimate, schedule effect, risk assessment, funding source, contractual implications and decision owner.

Change categories

A constructive change is especially dangerous because the supplier or team may perform additional work while the commercial record remains unclear. Prompt documentation protects both the client and provider.

6. Rebaselining: Discipline Rather Than Concealment

A baseline is a formally approved reference against which performance is measured. Rebaselining changes that reference. It can be legitimate after approved scope change, major authorized reset, contract modification or a fundamental planning assumption becoming invalid. It is not a method for erasing poor performance.

A sound rebaseline should:

• identify the approved change or trigger;
• preserve the historical baseline and variance record;
• reconcile budget, schedule and scope;
• update estimate assumptions and risk allowances transparently;
• maintain traceability from old to new values; and
• receive authority independent of the person seeking to conceal variance.

The project should distinguish a baseline change from a forecast change. A forecast may change every reporting period without changing the approved budget. Rebaselining should be infrequent and governed.

7. Corrective Decisions and Cost Recovery

Corrective action should address the cause, not merely the symptom. A structured decision sequence is:

1. detect and quantify the variance;
2. diagnose whether it arises from scope, estimating error, productivity, price, accounting timing, risk, supplier performance or unauthorized work;
3. test whether the cause is controllable, external, one-time or recurring;
4. develop alternatives and estimate their cost, benefit, risk and time effect;
5. select the least-cost action that protects required value and compliance;
6. assign an owner and measurable recovery criterion; and
7. monitor whether the action actually improves the forecast.

Common interventions include resequencing work, redesigning for constructability, changing suppliers, improving production methods, reducing rework, renegotiating terms, freezing discretionary scope, accelerating selected activities, or approving additional funding where value justifies it.

Cost recovery may arise from:

• contractual variations and approved change orders;
• claims for compensable delay or differing site conditions;
• supplier warranty or defect obligations;
• insurance recovery;
• tax, duty or incentive adjustments;
• productivity improvement; and
• value engineering that preserves required function at lower life-cycle cost.

Recovery must not be confused with shifting cost to another account. A claim should be supported by contemporaneous records, causation, entitlement, quantum and contractual notice requirements. Ethical practice prohibits inflating claims, misclassifying costs, delaying recognition of liabilities or presenting unapproved recovery as secured income.

8. Governance, Behaviour and Ethics

Forecasts influence funding, executive confidence, supplier negotiations and sometimes individual rewards. This creates incentives for optimism bias, strategic misrepresentation and “gaming” of EVM measures. Examples include deferring invoices, overstating completion, moving costs between work packages, keeping a weak baseline unchanged while informally changing scope, or labelling recurring variance as exceptional.

Governance controls should include:

• separation of forecasting, approval and assurance roles;
• documented assumptions and confidence ranges;
• independent estimate reviews at major gates;
• reconciliation between project controls and the general ledger;
• change authority thresholds;
• audit trails for baseline revisions; and
• escalation rules for material adverse trends.

A mature reporting culture treats bad news as decision information. The purpose of forecasting is not to defend the past but to improve the future outcome.

9. Technology and AI in Forecasting

Modern project-control platforms integrate schedules, cost ledgers, procurement data, timesheets, quantities and risk registers. Machine-learning models can identify patterns in invoice timing, productivity, rework, supplier behaviour and cost escalation. Natural-language tools can classify change requests and extract obligations from contracts.

However, AI forecasts require disciplined governance. Project data may be incomplete, biased or inconsistent; a model may learn historical practices that are no longer valid; and an apparently precise prediction may conceal uncertainty. Human experts remain responsible for validating causal assumptions, protecting confidential information, explaining decisions and approving changes. AI should support, not replace, accountable professional judgment.

10. Illustrative Case: MetroLink Station Package

Background

MetroLink is an illustrative light-rail station package with an approved budget of $48 million and a planned duration of 24 months. At month 14, the project has PV = $27 million, EV = $24 million and AC = $29 million. A design clarification, inflation in structural steel and repeated rework have affected performance.

Problem

The project sponsor must determine whether the original budget remains credible, whether a station-access change should be approved, and what recovery actions are justified.

Analysis

The current indicators are:

The CPI-only forecast is:

The forecast overrun is approximately $9.99 million. A bottom-up review identifies $2.2 million of one-time rework, $3.0 million of remaining steel escalation, $1.4 million in approved but not yet budgeted access requirements, and $18.0 million of remaining base work. If these values are complete and credible:

The bottom-up forecast is lower than the CPI-only forecast because some past variance is non-recurring. It is not automatically correct: the team must test whether rework controls have truly changed and whether the $18.0 million includes all commitments and residual risks.

Approach and solution

The change board approves the access requirement in principle because it is necessary for regulatory compliance and operational safety. It rejects an unrelated architectural enhancement because its benefits do not justify the cost and schedule impact. The approved change is funded through the authorized change-control reserve, with the baseline updated transparently.

Management implements a recovery plan: design freeze at defined interfaces, an independent constructability review, supplier expediting for critical steel, daily rework tracking, and a revised inspection hold-point process. A monthly bottom-up ETC is required for the three highest-risk work packages. The project reports a central EAC of $53.6 million, a downside scenario of $56.5 million, and a favourable scenario of $52.2 million, with assumptions explicitly stated.

Outcome and learning points

The case demonstrates why a mechanical CPI forecast should be challenged but not ignored. The CPI signal identified a serious problem; the bottom-up analysis explained its causes. The approved regulatory change was separated from poor historical performance, and the original baseline was preserved for accountability. Recovery actions were assigned to measurable causes rather than expressed as an unsupported demand to “work harder.”

Challenges and Limitations

Forecasting is limited by:

• incomplete or delayed actual-cost data;
• subjective measurement of physical progress;
• unstable scope and immature design;
• correlated risks that invalidate simple expected values;
• inflation, exchange-rate and supply-chain volatility;
• strategic behaviour and optimism bias;
• false precision in model outputs; and
• difficulties in attributing schedule effects to cost.

Managers should report ranges, confidence, assumptions and decision consequences—not merely a single number. Forecast quality should be assessed retrospectively by comparing prior forecasts with actual outcomes and identifying systematic bias.

Best-Practice Checklist

• Establish a measurable, authorized baseline.
• Reconcile actual costs, commitments and accruals before forecasting.
• Use EVM as an early-warning system, not as a complete diagnosis.
• Select EAC methods according to the cause and persistence of variance.
• Maintain bottom-up forecasts for material or changing work packages.
• Separate forecast changes from baseline changes.
• Evaluate every change for integrated scope, cost, schedule, risk and benefit effects.
• Preserve historical baselines and decision records.
• Use scenarios and sensitivity analysis when uncertainty is material.
• Link corrective actions to root causes, owners and measurable outcomes.
• Document claims and recovery with evidence and contractual entitlement.
• Ensure independent review and ethical escalation of adverse information.

Chapter Summary

Forecasting transforms current project evidence into an informed view of final cost. Earned value measures such as CV, CPI, EAC, VAC and TCPI provide useful signals, but their validity depends on reliable scope, progress and accounting data. EAC may be calculated through bottom-up estimating, continuation of current cost efficiency, combined cost and schedule efficiency, or an atypical-variance assumption; the selected method must match the project’s causal reality.

Trend analysis, scenarios, sensitivity analysis and simulation help managers understand uncertainty rather than conceal it. Integrated change control evaluates effects across scope, cost, schedule, quality, risk, contracts and benefits. Rebaselining is legitimate only when authorized and traceable; it must not erase historical accountability. Corrective decisions should address root causes, while cost recovery requires evidence, entitlement and ethical reporting. Effective governance combines quantitative controls, independent assurance, transparent assumptions and professional judgment.

Key Terms

Budget at completion: The total approved performance budget for defined project scope.

Actual cost: Cost incurred for work performed by a specified status date.

Planned value: Budgeted value of work scheduled by a status date.

Earned value: Budgeted value of work actually completed by a status date.

Cost variance: Difference between earned value and actual cost.

Cost performance index: Ratio of earned value to actual cost.

Estimate at completion: Forecast total cost at project completion.

Estimate to complete: Forecast additional cost required to finish remaining work.

Variance at completion: Difference between budget at completion and estimate at completion.

To-complete performance index: Efficiency required on remaining work to meet a target.

Trend analysis: Examination of performance behaviour across successive periods.

Scenario forecast: Forecast based on a coherent set of alternative assumptions.

Sensitivity analysis: Analysis of the effect of changing selected assumptions.

Integrated change control: Formal evaluation and authorization of project changes.

Baseline: Approved reference used to measure and control performance.

Rebaselining: Authorized revision of an established baseline.

Corrective action: Action intended to bring performance into conformity with the plan or requirement.

Cost recovery: Legitimate restoration or avoidance of project cost through contractual, operational or financial means.

Constructive change: An effective change caused by direction or conduct without formal change documentation.

Review Questions

Short Answer Questions

1. What is the difference between estimate at completion and estimate to complete?
2. Why should schedule performance not be interpreted solely from schedule variance in currency?
3. State two conditions under which a bottom-up ETC is preferable to a formula-based EAC.
4. What is the purpose of a change control board?
5. Why must historical baselines be preserved after rebaselining?

Long Answer Questions

1. Compare the major EAC forecasting methods and explain the assumptions underlying each.
2. Explain how trend analysis, scenario forecasting and sensitivity analysis complement EVM.
3. Describe a complete integrated change-control process from initiation to verification.
4. Discuss the ethical and governance risks associated with project cost forecasting.
5. Explain how corrective action and cost recovery can be integrated without disguising project overrun.

Analytical / Application Questions

1. A project has BAC of $20 million, EV of $9 million and AC of $11 million. Calculate CPI, CPI-based EAC, VAC and TCPI to meet BAC. Interpret the results.
2. A supplier claims that a design change caused a $700,000 overrun. Identify the evidence and analysis required before recommending recovery.
3. A project’s CPI has declined for four consecutive months, but the latest month includes a one-time equipment failure. Recommend a forecasting and corrective-decision approach, including the information you would request.

Multiple Choice Questions with Answers

Question 1

Which measure represents the forecast total cost of a project at completion?

A. Planned value

B. Actual cost

C. Estimate at completion

D. Earned value

Correct Answer: C. Estimate at completion

Question 2

A project has EV of $4 million and AC of $5 million. What is its CPI?

A. 0.80

B. 1.25

C. $1 million

D. −$1 million

Correct Answer: A. 0.80

Question 3

Which EAC method is most appropriate when past variance is clearly one-time and remaining work has been reassessed as valid?

A. AC plus remaining budgeted work

B. AC plus remaining work divided by CPI

C. AC plus remaining work divided by CPI times SPI

D. BAC multiplied by SPI

Correct Answer: A. AC plus remaining budgeted work

Question 4

What does a negative VAC generally indicate?

A. The project is forecast to finish under budget

B. The project is forecast to exceed its approved budget

C. The project has no remaining work

D. The baseline has automatically been approved for change

Correct Answer: B. The project is forecast to exceed its approved budget

Question 5

Which statement best distinguishes a forecast change from a baseline change?

A. A forecast may change without altering the approved budget

B. A forecast change always requires a new contract

C. A baseline change is made every reporting period

D. A forecast is only prepared after project completion

Correct Answer: A. A forecast may change without altering the approved budget

Question 6

What is the primary purpose of integrated change control?

A. To approve every requested change immediately

B. To evaluate total effects before authorizing changes

C. To eliminate the need for cost forecasts

D. To replace the project baseline with actual cost

Correct Answer: B. To evaluate total effects before authorizing changes

Question 7

If BAC is $12 million, EV is $6 million and AC is $7.5 million, what is the TCPI to meet BAC?

A. 0.80

B. 1.00

C. 1.20

D. 1.50

Correct Answer: C. 1.20

Question 8

Which practice is most consistent with ethical cost governance?

A. Deferring invoices to improve the current CPI

B. Reclassifying costs to hide a work-package overrun

C. Preserving the original baseline and documenting approved revisions

D. Reporting only the optimistic scenario

Correct Answer: C. Preserving the original baseline and documenting approved revisions

Question 9

What is a constructive change?

A. A completed defect correction with no scope effect

B. A change effectively caused by direction or conduct without formal documentation

C. A forecast prepared using three scenarios

D. A change to an accounting period

Correct Answer: B. A change effectively caused by direction or conduct without formal documentation

Question 10

Why should a CPI-based forecast be supplemented by a bottom-up review?

A. CPI can never be calculated from project data

B. Bottom-up review can explain whether current variance will persist

C. Bottom-up review automatically removes all risk

D. CPI measures schedule only

Correct Answer: B. Bottom-up review can explain whether current variance will persist