McKinsey 2015: The Source for "98% of Megaprojects Go Over Budget"
The most-cited megaproject overrun statistic on the web traces to one McKinsey article. We give its exact wording, the primary source, and the distinction between the 98% share and the 80% average that almost everyone merges.
Where does the "98% of megaprojects overrun" figure come from?
From McKinsey & Company's article Megaprojects: the good, the bad, and the better (July 2015). Its exact wording: 98% of megaprojects suffer cost overruns of more than 30%, and 77% are at least 40% late. The 80% figure quoted next to it is a different McKinsey number: an in-depth review of more than 300 billion-dollar-plus megaprojects that found average cost overruns of about 80% and schedule delays of about 50%. One is a share, the other is an average, and they should not be fused into "98% overrun by 80%".
The two figures, kept apart
The single most common mistake with this statistic is to quote it as "98% of megaprojects overrun by 80%". That fuses a count and an average that McKinsey states as separate findings. Here they are as the source presents them:
| Finding | Figure | What it measures |
|---|---|---|
| Megaprojects with a cost overrun above 30% | 98% | A share (how many overrun) |
| Megaprojects at least 40% late | 77% | A share (how many run late) |
| Average cost overrun, 300+ billion-dollar megaprojects | ~80% | A magnitude (how far over) |
| Average schedule delay, same review | ~50% | A magnitude (how far late) |
A project only has to slip just past 30% over budget to count toward the 98%. The ~80% average is a separate, harder number: across the 300-plus billion-dollar projects McKinsey reviewed directly, the mean overrun was roughly 80%, pulled up by a long tail of severe cases. Cite whichever you actually mean, and do not multiply one by the other.
The rounded "98% / 80%" shorthand is how the figure travels, and it is what our own construction and statistics pages use as a compact headline. This page is the long form: the two numbers, correctly separated, with the source attached.
What counts as a megaproject here
McKinsey's in-depth review covered billion-dollar-plus projects (contract values above 1 billion US dollars) across asset classes: transport, power, water, oil and gas, mining, and large buildings. That matches the $1 billion threshold Bent Flyvbjerg uses. The broader 98% / 77%-late headline is stated for megaprojects in general rather than for one named sample, which is why this page treats the 300-project review as the harder-edged, directly-attributable number.
McKinsey framed the article around a paradox it called the megaproject paradox: the world is spending more than ever on megaprojects (roughly 8% of global GDP at the time), while their delivery record barely improves. That is the same durable pattern Flyvbjerg calls the Iron Law of Megaprojects: over budget, over time, over and over again.
How it fits the wider evidence
The McKinsey megaproject figures are not an outlier. They line up with the independent, larger-sample datasets on this site:
- Flyvbjerg's 16,000-plus project database (2023): only 8.5% of projects come in on budget and on time, and just 0.5% also hit their benefits.
- Flyvbjerg, Holm & Buhl (2002): cost was underestimated in almost 9 in 10 transport infrastructure projects, by 28% on average.
- The Olympic Games: every Games since 1960 has exceeded its budget, averaging 157% cost overrun in real terms.
Different methods, different samples, the same direction. When McKinsey's 300-project review, Flyvbjerg's 16,000-project database, and the Olympic record all agree, the megaproject overrun problem is about as robust as an empirical finding gets.
How to cite
Source URL: mckinsey.com (megaprojects: the good, the bad, and the better)
The related McKinsey report The Construction Productivity Imperative (2015) carries the companion "large projects across asset classes typically take 20% longer to finish than scheduled and are up to 80% over budget" framing.