Cost Estimation vs. Actual Spending

Why Government Projects Go Over Budget and What Program Offices Can Do About It

Issue

Federal capital projects, from weapons systems and satellites to IT modernizations and infrastructure builds, routinely cost more than the estimates used to fund them. GAO has tracked this pattern for decades across Defense, NASA, the Department of Energy, and civilian IT programs, keeping acquisition management on its High-Risk List year after year. Overruns are not confined to a few troubled programs; they are a structural feature of how the government scopes, estimates, and executes large undertakings.

The root causes recur: optimistic technical and schedule assumptions made before a design is mature; historical analogies that don’t reflect a program’s actual complexity; underestimated integration and testing costs; requirements that grow after a baseline is set; and estimates shaped, deliberately or not, to compete favorably for funding rather than reflect real risk. GAO’s own best practices guidance identifies twelve characteristics of a high-quality cost estimate (comprehensive, well documented, accurate, credible) and finds most agency estimates fall short on several at once.

Impact

Cost estimating failures rarely stay contained within the program that produced them; the consequences ripple outward, compounding well beyond the program office that owns the original estimate:

• Program disruption: Cost growth forces rebaselining, descoping, or Nunn-McCurdy-style breach reporting, delaying delivery to warfighters, scientists, or the public by months or years.
• Opportunity cost: Every dollar absorbed by an overrun is a dollar unavailable elsewhere, letting the worst estimated programs crowd out well managed ones in a constrained budget.
• Erosion of trust: Repeated overruns weaken the credibility of future budget requests before Congress and OMB, inviting more rescissions, continuing resolutions, and micromanagement.
• Misaligned incentives: When breaching a cost baseline carries little consequence, managers and contractors have less reason to estimate conservatively or surface bad news early.
• Diminished mission outcomes: Descoped requirements and deferred maintenance show up later as capability gaps, cybersecurity debt, or safety risk, costs harder to trace back to the original estimating failure.

Because cost, schedule, and performance are interdependent, an estimate wrong on one dimension typically distorts the other two. A program that appears on budget may be achieving that only by quietly cutting scope or deferring testing. Cutting scope holds the budget line but delivers less capability than funded: cost held, performance quietly gave way. Deferring testing holds the schedule on paper while pushing defect discovery downstream, where fixes cost more and often force the very slip the deferral was meant to avoid. Either way, the estimate looked accurate only because performance or schedule absorbed the shortfall.

Action

Agencies do not need new statutory authorities to close most of this cost estimating gap between planned and actual program cost; the discipline already exists in GAO’s Cost Estimating and Assessment Guide and in DoD, NASA, and OMB Circular A-11 policy. What is missing is consistent application. Program and portfolio leaders should:

• Require independent cost estimates (ICEs) at each major milestone, performed separately from the program office, and reconcile them formally rather than defaulting to the more optimistic number.
• Ground estimates and planning in historical performance. Use historical cost, schedule, and contractor performance data from comparable programs, and insist on data driven assumptions, not optimistic ones, when setting budgets, schedules, and requirements.
• Align estimate maturity with design maturity and present viable options. Before preliminary design review, require risk-adjusted ranges and confidence intervals, not single-point estimates, supported by S-curves, sensitivity analysis, and alternative execution strategies so leaders can weigh realistic cost, schedule, scope, and performance tradeoffs.
• Fund and staff cost estimating functions as a core capability, not an afterthought; many overruns trace to estimating teams that were understaffed or brought in too late to influence key decisions.
• Treat the cost estimate as a living document. Keep the estimator informed of changes to requirements, scope, schedule, and technical assumptions, and maintain an auditable trail for every major assumption so the agency can trace and correct what failed.
• Report EVM data honestly and continuously: use cost and schedule performance indices to forecast Estimate at Completion (EAC) and act on early warning signs rather than waiting for a formal breach.
• Build management reserve and schedule margin into baselines deliberately, quantified through risk analysis, instead of treating contingency as padding to strip out during budget negotiations.
• Hold estimating teams and program managers accountable at post program reviews, and push reliability upstream to contractors: validate proposals against historical performance, require a documented basis of estimate and cost realism analysis during solicitation, and select contract types favoring fixed price where mature that place risk where it can best be managed.

None of these steps eliminate the uncertainty inherent in large, technically complex programs. But they replace optimism bias and one-time estimates with a disciplined, risk informed process, the one most consistently associated with programs that deliver within their approved baselines.

 

Here at Andrew Morgan we help organizations obtain the highest value from their human capital, organizational, and technical investments to lower costs and optimize how they work, with capabilities spanning business process improvement, systems engineering and integration, analytics and reporting, infrastructure modernization, regulatory risk/compliance, and security and information assurance. If you are looking for a partner to help navigate your organization or program through this new way of working, we would love to hear from and engage with you directly.

 

References

1. A Nunn-McCurdy breach is a statutory reporting requirement, named for the 1982 amendment sponsored by Senator Sam Nunn and Representative David McCurdy, that is triggered when a major defense acquisition program’s unit cost growth exceeds set thresholds, requiring the Department of Defense to notify Congress and, above a critical threshold, to certify the program’s continuation or terminate it.
2. U.S. Government Accountability Office, Cost Estimating and Assessment Guide: Best Practices for Developing and Managing Program Costs, GAO-20-195G (Washington, D.C.: March 2020).
3. Office of Management and Budget, Circular No. A-11, Preparation, Submission, and Execution of the Budget(Washington, D.C.: Executive Office of the President, updated annually).