Megaprojects are increasingly gaining worldwide popularity, often justified by claimed economies of scale and future expansion.
However, researchers at Oxford University now find that these propositions could be false, leaving large capital investments at risk to random events. Assuming bigger to be better “is a recipe for failure,” they warn in new research set for publication later this year.
“If you can avoid going big, do it, because it’s the most efficient way to avoid risk,” says Bent Flyvbjerg, professor of major program management at Oxford’s Saïd Business School and co-author of Big is Fragile: An Attempt at Theorizing Scale, to be published by the university. The better option is “to have big projects that are a combination of small projects,” he adds.
A giant wind-turbine farm is scalable and “fairly simple” compared to a nuclear or hydro power plant, he says. However, “a large hydro dam is binary. It’s either there, or it’s not. You can’t have it 90% finished,” he adds. In terms of cost overruns, his research shows “a very significant difference between those modular energy projects and the big, chunky ones.”
If a project’s “aspiration is scalability … and you come up with ‘big’ as the answer to fulfill that, there’s probably a flaw in the analysis,” adds Atif Ansar, a program director at the business school. The researchers define “scalability” as the ready ability to grow or shrink a project’s capacity in line with demand.
Big capital investments tend to be loss-making because of their long gestation, implementation and operation periods, asserts Flyvbjerg’s team. Any available economies of scale tend to be negated by the projects’ disproportionate exposure to uncertainties, they note.
As evidence, the team cites 245 large dams built in 65 countries between 1934 and 2007, together valued at $353 billion at 2010 prices. Cost overruns totaled 100% for one in five dams and 200% for one in 10. Extra costs prevented nearly half the projects from recovering their initial investment.
After construction, the risk of cumulative losses exceeding cumulative gains—also knows as “investment fragility”—increases with time because of processes, such as material fatigue. Further, costs of maintaining or removing obsolete installations represent a large potential liability. “Forecasts today are likely to be as wrong as they were between 1934-2007,” say the analysts.
To help explain the popularity of big projects, Flyvbjerg asserts that engineers—and people in general—are prone to “lock in,” whereby they focus on particular solutions without considering alternatives. “We like to focus on one thing quickly. It simplifies our lives … but it creates a lot of problems later, if you anchor on the wrong thing,” he says.
In Pakistan, for example, big dams are the accepted solution to the dual national challenges of electricity shortages and flooding, says Ansar. “But almost all other alternatives … are very readily ignored,” he adds. “Sometimes, experts are enlisted … for no other reason than they are experts in big dams.”
Even when engineers analyze a large project’s vulnerability to overruns, “they are looking at an outcome base that is way too narrow compared to what reality is throwing at us,” says Flyvbjerg.
It’s not enough to assess, for example, the impact of a 10% cost overrun on a project, he says. Project planners often say that “‘nobody ever does sensitivity analysis at 300% or 600% cost overrun,’” Flyvbjerg adds. “That’s the problem. Even when people look at history, they ignore [it]. They say, ‘It’s not going to happen this time.’ ”
Furthermore, when project proponents forecast costs, they “really mean the median of a probability distribution,” says Ansar. “What they forget to tell you is the overall variance and how wide is the tail of that distribution.”
These statistical terms define the likelihood of extreme outcomes, with a “fat tail” indicating higher probabilities, the Oxford researches explain.
“You can’t trust any median or average because it’s all about the tail,” says Flyvbjerg. Of the various types of projects, nuclear plants and hydro dams “have the highest cost overruns and the fattest tails, meaning the highest risks.”
One way of establishing realistic project forecasts is to review historical data “and identify what we call the black swans—the outliers,” says Flyvbjerg. “How did it happen? Could it happen again?” For a big dam, 10 to 15 historical samples should be reviewed, advises Ansar. “Look at their average cost overruns and more extreme overruns …and then stress test to really big events to see if the investments still pay off,” he notes.
Generally, decision-makers should treat data from project appraisals skeptically and require tough stress tests that reflect the full range of possible risks, advise the Oxford team. They caution against approving big projects on small benefit-cost margins. “Big investments,” they suggest, “need far more cushioning to avoid fragility than current management practice tends to assume.”