Tim Harford on failure
Tim Harford is a British economist, a columnist for the Financial Times and author of the new book, “Adapt: Why Success Always Starts With Failure.” We spoke last week about why governments have so much trouble failing well, whether voters should prefer indecisive politicians and what we can learn from the Iraq surge. Parts of this interview ended up in my column on the constructive role that failure can, and perhaps must, play in the health-care system.
Ezra Klein: The basic argument of your book is that the only way to solve complex problems is to fail toward the correct solution. But one of the things you suggest is that this is particularly hard for governments to do. Why?
Tim Harford: Let’s think about the balance of risks in the market or the scientific method. In both cases, you could have 50 failures and one success and you’ll still come out ahead. The theory of relativity and Google and penicillin more than make up for all the failed experiments, theories and businesses. The same is true, of course, of biological evolution. The number of failures are orders of magnitude larger than the successes.
Now think about politics. Any politician knows they can have 50 policies going well and one failure the failure will dominate the next campaign. So the politician is just desperate to avoid provable failure. And they can do that either by never doing anything or by refusing to quantify and evaluate what should happen when they do do something, as that way no one can prove it went wrong.
EK: One way this manifests, I think, is that there’s a preference for sweeping, blunt policies that may or may not work as opposed to more modest policies that accumulate over time but require patience and diligence. There’s something more satisfying about a big plan, like the Ryan budget, where you can just say it cuts costs by trillions of dollars by refusing to pay for more health care, than something like the Affordable Care Act, where it’s saying it’ll cut costs through aggressively experimenting with new ways to deliver care and hoping some of them will work.
TH: In politics, we like people who sound like they have a plan. Everyone remembers the Bush-Kerry election. And that election seemed to come down to one guy could make up his mind and one couldn’t. John Kerry’s defenders, as I recall, said he was decisive and that decisiveness just wasn’t coming through. Not that many tried to sell the idea that indecision is a virtue, that the world is uncertain and it makes sense to change your mind. In the U.K., Margaret Thatcher and Tony Blair won six elections, and she famously said, “you turn if you want to, the lady is not for turning.” And Blair, after British forces had gone into Iraq, he said, “I don’t have a reverse gear.” Imagine if I tried to sell you a car that didn’t turn or have a reverse gear. Would you want to buy it? Of course not. But we like those politicians.
EK: One of your chapters describes the turnaround in Iraq as a process of learning from trial-and-error that did happen in the U.S government. What were the lessons you drew from that for how governments can learn through failure?
HR: You could only loosely describe the U.S. army as a government body, of course. But it was the guys on the ground, the colonels and the majors, who were doing the innovating. I remember one guy said to me that though we learn at the bottom, they don’t learn at the top. Petraeus’s great strength was to assist and capitalize on the learning that was already taking place. That was one case where you could change because you absolutely had to change. The middle managers were watching their people get killed! But more conventionally, what it shows is that one thing that helps is if there’s natural variation anyway. The states are the laboratories of democracy, as they say. Someone tries something new and everyone else can have a look.
EK: I find it hard to mesh these ideas with the performance of health-care systems around the world. Your system in the United Kingdom is arguably the single most centralized, socialized system of any developed country. But costs are much lower, and quality not all that different, than in our system. Meanwhile, we have a lot more variation, but the outcome has been a disaster. How do you explain that?
HR: Generally, I don’t have much patience with people who worry about the waste and cost inherent in experimentation. Look at the Soviet Union. What looks to be efficient rationalization is often a system that can’t learn or adapt. But the question is whether health care an exception to that. The system in the U.K. is centralized. But there’s still a tremendous amount of trial-and-error built into it. It’s medicine, after all. You have randomized trials going on constantly. You have the journals of medicine. I wonder if they’re able to get away with a bureaucratic centralization because the actual medical practice remains very experimental.
EK: One possibility that your answer seems to suggest is that the useful site for experimentation in the health-care system is in medical treatments, not insurance products. What we really have is enormous variation in insurance products and arrangements and that seems to have mostly led to a lot of cost.
TH: My guess is that insurance in the U.S. system is an important mediator of the whole thing costing a tremendous amount. There are some things you can predict on theory alone but a system where the costs are shared but the choices are made by individuals is a system where you’d expect an awful lot of unnecessary or marginal health care. I remember an article by Malcolm Gladwell on this subject, where he said no one goes and has a hip replacement for fun. I thought he was wrong. There are many procedures done because they are effectively free or very cheap for the person who decides whether to have them done, profitable for the person who decides whether the procedure should happen, and the bill is paid elsewhere.
EK: My understanding, though, is that Americans aren’t really overtreated as compared to Europeans. In many cases, we get less care than you guys do, but pay much more per unit of care. I’ve always thought the strongest argument in favor of these costs is that we’re funding innovation that the rest of the world free rides off of. On the other hand, simply overpaying for things is probably not the smartest way to goose innovation.
TH: I think that’s plausible. There probably is more innovation that you guys are paying for, so thanks very much, we all owe you.