Earned value management: the three measures and two variances

A cost report holds two series: what was planned to be spent, and what has been spent. Both answer the same question, how much has been paid, and neither answers the one that matters on a running project, how much has been produced. The standard that governs the method says so itself: without earned value, one can only compare planned expenditures with how much has been spent, which does not provide an objective indication of how much of the planned work was actually accomplished (SAE EIA 748E, Revision E, 2026, scope). On those two series alone, a work package spending less than planned and a work package barely moving look the same.

What earned value is

In traditional management there are two data sources, the budget or planned expenditures and the actual expenditures. In earned value management there are three (DOE, Earned Value Management Tutorial, Module 1, 2003). That is the whole of the change, and it is worth saying early, because it is smaller than the reputation of the method suggests. Two of the three measures sit in every cost report already. One is new.

The two measures every cost report already holds

Planned value is the time-phased budget authorised for the work scheduled, also known as budgeted cost of work scheduled (ISO 21508:2018, 3.1.14). It is not a new object: it is the cost baseline, read as a curve over time. How that baseline is built, what sits inside it and who may change it belongs to the article on contingency and management reserve, and nothing here reopens it.

Actual cost is the cost incurred for work performed (same standard, 3.1.2). Of the three measures it is the only one accounting supplies with no additional work.

Both definitions are cited here from ISO 21508:2018, withdrawn in February 2026: since its 2026 edition, ISO 21508 no longer carries the terms and refers to ISO 21506 for them (ISO 21508:2026).

The third measure, earned value

Earned value is the value of completed work expressed in terms of the budget assigned to that work (ISO 21506:2024, 3.21). Its admitted term is budgeted cost of work performed, and why one quantity carries two names is the subject of a later section.

The verb is not a metaphor, it is the mechanism. A target planned value, that is, a budget, is established for each scheduled element of work, and as these elements of work are completed, their target planned values are earned (SAE EIA 748E, Revision E, 2026, scope).

How that completion is established is a separate question, and a prior one: the measure is agreed before the work package starts and does not change once work has begun (ISO 21508:2018, 5.6). This article takes the measurement as given and puts a figure in money on it. Which crediting rule applies, and how it is chosen, belongs to physical percent complete.

One property separates the third measure from the second in kind, and most introductions pass over it. Earned value is bounded: the earned value of a completed work package always equals its budget (ISO 21508:2018, 5.6). Actual cost is bounded by nothing, its definition being simply the cost incurred for the work performed. A finished work package has earned exactly its budget, whatever it cost to get there. Set the two definitions side by side and the asymmetry follows. No standard in our set states it as such, and it is the whole reason the third measure is not a duplicate of the second.

The two variances, and the two indices

With the third measure available, the two useful comparisons write themselves, and both start from the same point. Cost variance is earned value minus actual cost. Schedule variance is earned value minus planned value (ISO 21508:2018, Annex A, Table A.1). The first sets what the work is worth against what it cost. The second sets the work done against the work planned.

The indices are those same two comparisons expressed as ratios rather than differences: the cost performance index is earned value divided by actual cost, the schedule performance index is earned value divided by planned value (same annex). At this level their only contribution is to make work packages of different sizes comparable with each other.

One reading of that chart is counter-intuitive enough to be worth stating on its own. Schedule variance carries the word schedule and is expressed in the unit of the budget. It compares an amount of work with an amount of work, never a date with a date. A schedule variance of minus 20,000 euros does not say how many days late anything is. ISO 21508:2018 puts the limit in the same annex: the earned value schedule measures are cost measures, and their reading has to be confirmed by network analysis, which remains the primary source of time-based information.

The wording came out that way at the start. The original text defines schedule variance as earned value minus budget (ANSI/EIA-748-1998, 2.6), the word budget standing where planned value stands today. The quantity has not moved. Only its name has.

What it is for on a contract with a fixed term

Where the end date and the price are both fixed, the value of the method is not the measurement itself. It is the date at which a problem becomes visible. An overrun readable at a third of the work accomplished can still be acted on. The same overrun established at delivery cannot.

That is what the instrument has actually produced, including on programmes where nobody acted on it. Reviewing the A-12 Avenger and the C-17, Wayne F. Abba wrote that earned value analysis revealed that the problems were apparent, if not avoidable, long before the contractors and program managers acknowledged them (PMI, How earned value got to primetime, 2000).

A working system carries more than the two variances. It also carries the analysis of significant variances and the forecast at completion, which is usually what the buying organisation is actually waiting for (SAE EIA 748E, Revision E, 2026; NDIA, EIA-748-E Intent Guide, Revision E, version of 14 May 2026). Forecasting is a subject of its own and is left aside here.

A worked example

The figures below are invented, and round on purpose.

A work package is budgeted at 100,000 euros. At the status date, the planned value is 60,000 euros, the measured progress is 40 percent, and the actual cost is 52,000 euros.

• Earned value: 40 percent of 100,000, so 40,000 euros.
• Cost variance: 40,000 minus 52,000, so minus 12,000 euros. Cost performance index: 40,000 divided by 52,000, so 0.77.
• Schedule variance: 40,000 minus 60,000, so minus 20,000 euros. Schedule performance index: 40,000 divided by 60,000, so 0.67.

Read on the two series of an ordinary cost report, this work package had spent 52,000 where 60,000 were planned. It was under budget, and it was reassuring. The three measures say the opposite, twice: the work accomplished is worth less than it cost, and less than what was planned for this date.

The 40 percent is an input here, not a result. Where it comes from is the crediting rule, settled before the work package starts and not revisited in this article.

Where it comes from

The problem: a baseline that moved

Earned value was not born out of a need to measure. It was born out of a need to stop the baseline from moving.

Its competitor at the time allowed the baseline to be adjusted each month so that the sum of the actual costs and the estimate to complete would equal the original budget, masking overruns until the money ran out. James B. Morin, writing its history, calls that the ultimate flexible baseline (How It All Began, 2016, p. 16). Earned value made the opposite choice: standards derived from the subdivision of the contract, with change controls limiting any revision to the addition or deletion of specific work. A non-flexible baseline.

The immediate motive was narrower still. In 1965, estimates at completion were notoriously inaccurate, and finding a solution to that problem was, in Morin's words, the primary objective of earned value (same source).

This is what makes the history worth a section rather than an ornament. The rigidity of the baseline, which practitioners meet today as a constraint, is the reason the method exists. It is not a side effect of it.

Three generations of acronyms, one set of measures

Anyone reading around the subject meets several sets of initials for the same three measures. This is not a defect in any one document. It is sedimentation.

The first set dates from 1966. Commenting on the specification at first hand, Lorette and Roth describe a comparison of the planned value of work scheduled, the planned value of work accomplished, and the actual costs incurred for the cost account (Lorette and Roth, 1970). The second set, BCWS, BCWP and ACWP, came with the contractual criteria of the following decade. The third, PV, EV and AC, is the modern vocabulary and the one ISO uses.

The decisive fact is not the succession, it is the coexistence. Two of these sets are in service today inside the same administration. The Earned Value Management Gold Card published by the DAU uses BCWS, BCWP and ACWP and nothing else (2020). The DOE EVMS Gold Card prints both, PV or BCWS, EV or BCWP, AC or ACWP (version of 8 March 2019). Two practitioners holding one card each are reading the same quantities under different names, and neither document is out of date. That is why the confusion does not clear on its own.

French-language documents carry sets of their own on top of these, the Dictionnaire de management de projet using CBTP, CBTE and CRTE (AFNOR and AFITEP, 3rd edition, 1996), which is why a translated glossary rarely settles the question either.

A definition unchanged since 1998

ANSI/EIA-748-1998 defines earned value at clause 2.6 as the value of completed work expressed in terms of the budget assigned to that work. ISO 21506:2024 defines it at clause 3.21 as the value of completed work expressed in terms of the budget assigned to that work. Twenty-eight years, two standards bodies, and not one word between them.

That is the useful conclusion of this section, and it closes the question the previous one opens. The names move, the quantity does not. What a reader takes from this article will not be made obsolete by the next edition of anything.

Who created the method is a different matter, and the literature does not agree. Morin credits A. Ernest Fitzgerald at Performance Technology Corporation in 1965 (2016). Abba credits the pioneers led by Hans Driessnack and does not mention Fitzgerald once (2000). Fleming and Koppelman trace the idea to the American factory floor of the late nineteenth century, and then to PERT/Cost (Earned Value Project Management, 2nd edition, 2000). All three accounts were published by the PMI or the College of Performance Management, and two of the three authors were directly involved in the events they describe. The disagreement is recorded here and not settled.

One dated fact on how the method was received: through the 1990s, responsibility for earned value in both government and industry was typically assigned to program control specialists and viewed as a financial reporting requirement (Abba, 2000).

In short, the missing input is rarely the money

A cost curve that cannot separate a late project from an expensive one is not short of a data point. It is short of a quantity.

The step that follows costs nothing and needs no tool. Take a work package currently under way and ask what would be required to compute its earned value. The budget is there. The actual cost is there. What is almost always missing is the third input: a rule, agreed in advance, for deciding how much of that work package counts as done.

That rule has to be chosen before the work package starts, it cannot change once work has begun, and there is only one of them per work package (ISO 21508:2018, 5.6). The intent guide states the same requirement from the other end: performance must be claimed in the same manner it was planned (NDIA, EIA-748-E Intent Guide, Revision E, version of 14 May 2026, guideline 14). An earned value figure assembled after the fact measures nothing at all.

Stop guessing. See the real impact.

Frequently asked questions

Q.What is the difference between earned value and physical percent complete?

Physical percent complete is a percentage, earned value an amount. The second follows from applying the first to the budget of the work package, an identity stated as such in the Dictionnaire de management de projet (AFNOR and AFITEP, 3rd edition, 1996). Measuring the percentage is a separate subject.

Q.Does earned value say anything about the critical path?

No, and the confusion is an expensive one. A schedule performance index aggregates volumes of work across a whole account: it names no activity, and it cannot say whether the shortfall sits on a critical activity or on one that still has float. Annex A of ISO 21508:2018 sends the question to network analysis for exactly that reason.

Q.Is a full earned value management system needed to use earned value?

No. The standard separates its principles, described as fundamental to all programs, from its guidelines, described as applicable to large, complex or high-risk programs (SAE EIA 748E, Revision E, 2026). Earned value as a quantity does not require the whole contractual apparatus.

References

• AFNOR, AFITEP - Dictionnaire de management de projet français-anglais-espagnol - 3e édition, 1996
• College of Performance Management - James B. Morin - How It All Began - The Creation of Earned Value and the Evolution of C/SPCS and C/SCSC - 2016
• DAU - Earned Value Management Gold Card - 2020
• DOE - Earned Value Management Tutorial Module 1: Introduction to Earned Value Management - 2003
• DOE - DOE EVMS Gold Card - Earned Value Management System Basics - Version of 8 March 2019
• Electronic Industries Alliance - ANSI/EIA-748-1998 - Earned Value Management Systems - Initial edition, approved 19 May 1998, published June 1998
• ISO - ISO 21506:2024 - Project, programme and portfolio management, Vocabulary - 1st edition, 2024
• ISO - ISO 21508:2026 - Project, programme and portfolio management - Earned value management - 2nd edition, 2026
• ISO - ISO 21508:2018 - Earned value management in project and programme management - 2018
• NDIA - EIA-748-E Intent Guide - Earned Value Management Systems - Revision E, version of 14 May 2026
• PMI - Wayne F. Abba - How earned value got to primetime: a short look back and glance ahead - 2000
• PMI - Quentin W. Fleming, Joel M. Koppelman - Earned Value Project Management - 2nd edition, 2000
• SAE International - SAE EIA 748E - Earned Value Management Systems - Revision E, 2026
• United States Air Force - Richard J. Lorette, Berton J. Roth - Cost/Schedule Planning Control Specification - 1970