In the last few months, the globalized world has endured two very different crises. First, there was the ash cloud over Europe, which paralyzed air travel for millions of passengers. Then, there is the leaking oil well in the Gulf of Mexico, which continues to spew somewhere between 5000 and 60,000 barrels of crude into the ocean every day.
While these disasters have nothing in common, our response has been plagued by the same fundamental problem. In both instances, officials settled on an early version of events – the ash cloud posed a severe danger to plane engines, and the oil well wasn’t a very bad leak – and then failed to update that version in light of new evidence. As a result, valuable time was squandered.
This is a form of anchoring, a mental bias first outlined (of course) by Kahneman and Tversky. One of my favorite demonstrations of anchoring was done by a group of MIT economists led by Dan Ariely, as they conducted an auction with their business graduate students. (The study was later repeated with executives at the MIT Executive Education Program with similar results.) The items for sale included everything from a fancy bottle of French wine to a cordless keyboard to a box of chocolate truffles. The auction, however, came with a twist: Before the students could bid, they were asked to write down the last two digits of their social security number. Then, they were supposed to say whether or not they would be willing to pay that numerical amount for each of the products. For instance, if the last two digits of their social security number were 55, then they’d have to decide whether or not the bottle of wine or the cordless keyboard were worth $55. Finally, the students were instructed to write down the maximum amount they were willing to pay for the various items.
If people were perfectly rational, then writing down their social security numbers should have no effect on their bids. In other words, a student with a low valued social security number (like 10) should be willing to pay roughly the same price as someone with a high valued number (like 90). But that’s not what happened. Look, for instance, at the bidding for the cordless keyboard. Students with the highest social security numbers (80-99) made an average bid of $56. In contrast, the average bid made by students with the lowest numbers (1-20) was a paltry $16. A similar trend held for every single item. On average, students with higher numbers were willing to spend 300 percent more than those with low numbers.
What does this have to do with the ash cloud and the oil spill? While anchoring is typically interpreted as a consumer mistake – we anchor to the wrong price – I think the bias can also be applied to our beliefs. Consider the ash cloud: After the cloud began drifting south, into the crowded airspace of Western Europe, officials did the prudent thing and canceled all flights. They wanted to avoid a repeat of the near crash of a Boeing 747 in 1989, when the KLM aircraft flew, at night, above the ash cloud of an Alaskan volcano. This led to a complete engine failure and an emergency landing.
Given the limited amount of information, anchoring to this previous event (and trying to avoid a worst case scenario) was the only reasonable reaction. The problems began, however, when these initial beliefs about the risk of the ash cloud proved resistant to subsequent updates. Over the next few days, numerous test flights were sent up into the atmosphere. Much of the data collected by these flights suggested that the ash might not be such a severe danger, at least at the periphery. For instance, measurements of the atmospheric ash concentration found that the density of particles was roughly comparable to the sand dust clouds from the Sahara that sometimes sweep over central Europe. (Surprisingly, the highest concentration of particles was found at the atmospheric boundary layer, which is below 3 km. However, these particles are mostly air pollution from cars and factories.) Furthermore, subsequent tests also revealed that the silica content of the Icelandic ash was significantly below that of other dangerous ash clouds, such as that produced by the Alaska volcano in 1989. (For a bunch of reasons, clouds of lower silica content are less likely to melt in engines at cruising speed.) All of this evidence helps explain why numerous test flights carried out by Lufthansa, Air Berlin and British Air during the crisis revealed no damage to the aircraft and aircraft engines. Nevertheless, the cancelations continued for another four days.
My point is absolutely not that the ash cloud wasn’t dangerous, or that the aviation agencies were wrong to cancel thousands of flights, at least initially. (Nobody knows the answer to those questions, although I’d rather be an inconvenienced traveler than a victim of catastrophic engine failure.) Instead, I think we simply need to be more aware that our initial beliefs about a crisis – those opinions that are most shrouded in ignorance and uncertainty – will exert an irrational influence on our subsequent actions, even after we have more (and more reliable) information. The end result is a kind of epistemic stubbornness, in which we’re irrationally anchored to an outmoded assumption.
The same thing happened with the BP oil spill. Initial reports suggested that the leak was relatively minor, less than a thousand barrels a day. As a result, BP and the government were slow to launch into crisis mode. Days were squandered; little was done. In fact, it took BP seventeen days before the company attempted the first fix of the underwater leak. Because BP and the government were anchored to a false belief – the spill wasn’t supposed to be that bad – they spent weeks thinking the wrong thing in a crisis.
The only way to avoid anchoring is to know about it. We need to be more aware that anchoring is a fundamental flaw of human decision making, and that our first reaction to an event will continue to shape our ensuing thoughts, even after that reaction is no longer relevant or valid. Our old beliefs might be wrong, but they’re influence lingers on, an intellectual anchor holding us back.