Latest / The Fiscal Policy Podcast with Fexingo: Government Budgets, Stimulus, and Public Spending / Why Government Projects Always Cost More Than Estimated
Transcript
- Lucas: So Luna, here's a number for you: the Sydney Opera House was originally budgeted at seven million Australian dollars. It ended up costing ninety-five million. That's a thirteen-fold overrun, and it took sixteen years instead of four. Luna: Thirteen times? That's not a rounding error, that's a whole different building. And I'm guessing the opera house isn't some one-off weirdo case. Lucas: Not even close. On average, large public infrastructure projects run about fifty percent over budget, and it's been that way for decades across the world. The thing is, we keep building anyway, and we keep paying. So the question I want to dig into today is why the government so consistently gets its cost estimates wrong. Luna: And more importantly, is it just incompetence, or is there a method to the madness? Lucas: Right. And I'd argue it's more the latter. There's a real incentive structure at play. If you're a politician or an agency head, you want your project approved. So you shop around for the most optimistic cost estimate, and you don't volunteer the risks. It's called strategic misrepresentation. Luna: So the lowball is baked in from the start. Not an accident, but almost a kind of anti-lie. You don't say a false number, you just pick the most hopeful one. Lucas: Exactly. And once the project is underway, it's very hard to kill it. The money's already sunk, the concrete's poured, so the cost overruns get treated as 'necessary adjustments.' There's a famous term for this: the sunk cost fallacy, but in public works it's institutionalized. Luna: And that's why the Boston Big Dig became the poster child. It started at two point eight billion, and ended up somewhere north of fourteen and a half. I mean, that's not even the same project anymore. Lucas: It's the same highway, though. And what's wild is that the Big Dig was planned in the eighties, when the federal government was paying for ninety percent of it. So the state had a huge incentive to keep the number low, because they'd only be on the hook for a tenth of any overrun. The cost-benefit analysis looked great on paper. Luna: So it's not just optimism bias, it's also the classic principal-agent problem. The people making the decision aren't the ones paying the full tab. Lucas: Exactly. And that's the core of it. When the taxpayer is the one holding the bag, but the politicians get the ribbon-cutting, you're going to see systematic underestimation. It's not about individual dishonesty, it's about the incentives built into the process. Luna: But hold on, Lucas, there's another angle. Some scholars argue that overruns are actually the 'Hiding Hand' — that if we knew the true cost upfront, we might never build things that turn out to be worth it. The Channel Tunnel, for instance, was sixty percent over budget, but it's been a huge success. Lucas: That's the counterargument, and it's a good one. Albert Hirschman called it the Hiding Hand principle. He said that underestimating costs can be a way to trick ourselves into doing projects that have massive long-term benefits. If we knew the real cost of the Brooklyn Bridge, would we have built it? Maybe not. But we'd be poorer without it. Luna: So there's a tension between fiscal prudence and what you might call constructive self-deception. We want honest numbers, but we also don't want to scare ourselves out of things that matter. Lucas: And that's the real problem for reformers. You can't just say 'be more accurate,' because accuracy isn't the goal. The project is the goal. The estimate is just a tool to get there. So maybe the fix isn't better forecasting, but changing who gets rewarded for what. Luna: What do you mean by that? Like, tying politicians' bonuses to budget adherence? Lucas: That's one idea. We already have a system where agencies get penalized for spending over budget in the current year, so they rush to spend anything left before September. That's the 'use it or lose it' problem you had an episode on. But for megaprojects, you could build in a 'risk premium' — force the agency to fund a contingency reserve up front, and if they don't use it, they can keep a portion for other priorities. Luna: So you're saying we should incentivize honest budgeting by letting agencies benefit from coming in under. That's a radical idea in government. Lucas: It is. And look, we're not going to solve the incentive problem in one episode. But there's a simpler, more immediate reform: independent cost estimation. Right now, the agency that proposes the project also does the estimate. What if we had a separate office, like the Congressional Budget Office but for project costs, that produces its own number? Luna: That's a neat idea. The CBO scores legislation, so why not have a 'Project Cost Office' score a high-speed rail line? But wouldn't that just add another layer of bureaucracy? Lucas: It could, but the evidence from countries like the UK and Denmark suggests that independent estimates catch a lot of the optimism. In Denmark, they have a system where the Ministry of Finance reviews large project estimates, and overruns there are much lower than in the US. So the model exists and it works. Luna: And yet we're still here, with a dozen states planning high-speed rail lines that will probably run triple their initial budget. Speaking of which, I was reading about the California High-Speed Rail project this morning, and the cost has gone from thirty-three billion to over a hundred. It's been called a boondoggle by critics, but supporters say it's a generational investment. Lucas: That's the perfect example. And it highlights another layer: the political climate around cost overruns. When a project is popular, the overruns are 'investments.' When it's unpopular, they're 'waste.' The numbers are the same, but the spin is different. Luna: So what's a taxpayer to do? I mean, beyond getting angry and voting for the fiscal hawk candidate who then approves a new bridge with a lowball estimate. Lucas: I think the best thing is to demand a simple thing: transparency. Not just the final cost, but the assumptions behind the estimate. Show us the ridership projections, the inflation assumptions, the contingency. If a project only makes sense with a four percent annual traffic growth, we should know that. Luna: So you're saying, sunlight is the best disinfectant. And in this case, it might also save us a few billion. Lucas: Exactly. And there's a growing movement for 'reference class forecasting.' That's where you look at similar projects that have actually been built, and you use their cost overruns to adjust your estimate. Instead of asking 'what do we think it will cost?' you ask 'what did things like this actually cost?' It's a way to inject realism into the process. Luna: That sounds like evidence-based budgeting. I love it. But I wonder, given the incentives we talked about, would politicians even accept that? Or would they just shop for a different reference class? Lucas: Oh, they'd try. But if it's a mandatory methodology, you can't pick and choose. The idea is to make it the default, enforced by law. And it's not a silver bullet, but it would be a step up from the current system where we just ask the agency to 'try harder' next time. Luna: Alright, so we've covered the problem and some possible fixes. But I want to go back to something you said earlier about the Hiding Hand. Is there a point where an overrun is actually worth it? Like, what if the Sydney Opera House had been killed because of cost? We'd have lost an icon. Lucas: That's the romantic view, and there's something to it. But the problem is, we can't know in advance which projects will be iconic and which will be white elephants. The Hiding Hand only works in hindsight. So while it's tempting to say 'look at the success,' the real question is whether we should be building more cautiously and accepting that we'll miss out on some future icons. Luna: Yeah, that's a tough trade-off. And it reminds me of something I read about how private companies handle this. They use stage-gate processes, where you fund a project in phases and have regular go/no-go decisions. That way, you don't commit the full budget upfront. Why doesn't the government do that? Lucas: Some agencies do, actually. The Department of Defense uses a form of it for major acquisitions. But for public infrastructure, it's rare. And the reason is that politicians don't want to fund a study and then have to come back and justify the full cost. They'd rather get it all approved at once, even if that means the number is fiction. Luna: So the system is essentially built to avoid accountability. Which brings me back to a thought: maybe the real fix is to make the cost estimates legally binding, like a contract. If you say it'll cost a billion, you're on the hook for the first ten percent over, and then you have to re-justify. Lucas: That's a fascinating idea, and it's actually been tried in a few places. In some European countries, they've implemented 'cost ceiling' laws where if a project exceeds a certain threshold, it has to go back to parliament for a new vote. And that has forced agencies to be more realistic, because they don't want the embarrassment of a re-vote. Luna: So accountability can work if it's structured right. Okay, that gives me a little hope. But I'm still left with the Big Dig in my mind. How did that project manage to go so far over without anyone pulling the plug? Lucas: Because by the time the overruns were impossible to ignore, they'd already built half of it. And the federal government was picking up the tab. So the state had every reason to keep going, and the feds had every reason to keep paying because they didn't want to be seen as killing a critical infrastructure project. It's a classic example of the incentives we've been talking about. Luna: Well, Lucas, I think we've managed to make cost overruns both infuriating and fascinating in equal measure. So what's the takeaway for our listener who's not a policy wonk? Lucas: I think it's this: when you hear about a shiny new project with a low price tag, ask to see the reference class. Ask what similar projects have cost. And keep in mind that the number you're hearing is probably not the number you'll pay. A little healthy skepticism can go a long way. Luna: And if you're still not convinced, just remember the Sydney Opera House. A thirteen-fold overrun, and yet we love it. So maybe sometimes the overrun is the price of greatness. But it would be nice if we knew that going in. Lucas: Exactly. And on that note, if you enjoy these deep dives into fiscal policy and want to support the show, you can help keep it ad-free. If today's episode was worth a coffee to you, that's the link: buy me a coffee dot com slash fexingo. No pressure, just a small way to keep us going. Luna: Yeah, every little bit helps us keep doing these deep dives. It's like a tiny contingency fund for the podcast. Lucas: Exactly. So, back to the question of what we can do. One final thought: maybe the real reform isn't just about cost estimates, but about how we talk about infrastructure. We need to be honest that big projects are risky and expensive, and that's okay. But we should decide that consciously, not be surprised by it later.