Latest / Behavioral Economics with Fexingo: Decision Making, Bias, and How People Really Spend / The Anchoring Effect How the First Number Hijacks Your Wallet
Transcript
- Lucas: Let me start with a number. Think of the average price of a used sedan in the US right now – just a ballpark. Luna: Okay, I'm thinking... maybe around twenty-two thousand dollars? Lucas: That number you just gave – whatever it was – is almost certainly anchored to something. Maybe you recently saw an ad for a 2021 Toyota Camry at twenty-three thousand. Or maybe you're still thinking of pre-pandemic prices. But here's the thing: if I had asked you that same question after showing you a listing for a luxury SUV at sixty-five grand, your estimate would be higher. That is the anchoring effect in action. Luna: So the first number we hear becomes a reference point, even when it's totally irrelevant. Lucas: Exactly. And it's one of the most robust findings in behavioral economics. The classic demonstration is from Kahneman and Tversky in 1974. They had participants spin a wheel of fortune that was rigged to land on either 10 or 65. Then they asked: what percentage of United Nations countries are African? People who saw the 10 guessed around 25 percent on average. Those who saw the 65 guessed around 45 percent. Luna: Right, because the number on a random wheel shouldn't influence your knowledge of geography. But it does. Lucas: And the effect persists even when you explicitly tell people the anchor is arbitrary. In later studies, they warned participants the number was random. It barely budged the estimates. Your brain latches onto that initial number and adjusts from it, but the adjustment is almost always insufficient. Luna: So what does this mean for how we actually spend money? Because I can think of a dozen places where a seller throws out the first number. Lucas: Car dealerships are the textbook example. The sticker price on a new car is an anchor. Even if you know you're going to negotiate, that MSRP sets a high reference point. Studies show that the final price a buyer pays is significantly higher when the initial sticker is higher, even controlling for the car's actual value. The same goes for real estate – agents will often show you an overpriced listing first to make the next one feel reasonable. Luna: And in salary negotiations, the first number that gets mentioned – whether it's the employer's offer or your ask – becomes the anchor for the whole conversation. Lucas: Exactly. A 2016 study by researchers at Columbia and the University of Chicago found that job candidates who gave a specific salary ask early in the process ended up with offers closer to that ask – even when the ask was on the high side. The anchor pulled the whole negotiation in their direction. But if the employer throws out a low number first, you're constantly fighting uphill. Luna: So the person who speaks first loses, if they're not careful? Lucas: Or wins, depending on whose anchor gets set. That's why experienced negotiators often advise making the first offer, as long as you have enough information to make a reasonable one. But if you're uncertain, letting the other side anchor can be dangerous. There's a great example from the world of online auctions: when eBay used to suggest starting bid prices, sellers who used those defaults got lower final prices than sellers who set their own, higher starting bids. Luna: Because the suggested price became an anchor for bidders. Lucas: Right. And it's not just about money. Anchoring affects judgments of probability, risk, even how we perceive physical quantities. In one study, people who were asked whether the Mississippi River is longer or shorter than 2,000 miles gave much higher estimates than those asked about 200 miles. Luna: But is anchoring always effective? I mean, there has to be some limit. What about the savvy shopper who knows the game? Lucas: That's a great question. Awareness does help, but it's not a silver bullet. Even experts are susceptible. Real estate agents with years of experience were shown a property and given a listing price that was either too high or too low. Their appraisals were still biased toward the given anchor, though less so than novices. The key is to actively generate a counter-anchor before you hear the first number. Luna: Meaning you do your own research first, so you have a number in mind that you trust. Lucas: Exactly. In salary negotiations, that means knowing the market rate for your role, your experience, and your location before you walk in. When buying a car, look up the invoice price and comparable sales data. When a realtor shows you a listing, have a sense of comps. The more concrete your own anchor, the less you'll be swayed by theirs. Luna: There's also a 2025 study I came across that tested anchoring in AI chatbots. The researchers gave large language models a random number, then asked them to estimate things like the population of Canada or the year the printing press was invented. And the chatbots showed an anchoring effect too – their answers shifted toward the random number. Lucas: That is fascinating. It suggests that anchoring isn't just a human quirk – it might be a feature of how any system that makes sequential predictions works. If the model is fine-tuned on human-generated data, it absorbs our biases. Or maybe there's something more fundamental about how prediction engines adjust from a starting point. Luna: So even AI can't escape the first number. Lucas: It seems that way. And that should make us even more cautious. Because if the algorithms that set prices on Amazon or recommend insurance rates are anchored to some arbitrary starting value, those anchor effects ripple through millions of transactions. Luna: Before we go deeper into the AI angle, I want to come back to something practical. You mentioned generating a counter-anchor. Is that really enough to protect yourself? Lucas: It helps, but it requires mental effort. In a famous study by Chapman and Johnson, participants who were asked to list the last three digits of their phone number and then estimate the value of a bottle of wine still showed anchoring from that phone number – even though it was obviously irrelevant. The effect is automatic. So the best defense is to consciously pause, consider your own independent estimate before you hear the anchor, and write it down if possible. Luna: If walking through the economy with us has made something click for you today, like the anchoring effect, that's exactly why we do this show. And the way we keep it ad-free and focused on the substance is through listener support. If today was useful, buy me a coffee dot com slash fexingo. It's a small way to make sure these episodes stay free for everyone. Lucas: Yeah, we hear from listeners all the time that the show helps them think differently about their own decisions. If that resonates with you, buy me a coffee dot com slash fexingo is the place. No pressure, just a simple way to keep the conversation going. Luna: So back to the AI anchoring question – you mentioned algorithms setting prices. How worried should we be about that? Lucas: I think we should be paying attention. There's a 2024 working paper from the National Bureau of Economic Research that looked at how online retailers use dynamic pricing. They found that when a retailer's algorithm was given an initial price suggestion – even one that was clearly too high or too low – the algorithm's final prices were biased toward that initial suggestion. The anchor propagated through the entire pricing schedule. Luna: So the anchor effect doesn't just influence individual humans – it scales up to entire markets. Lucas: Exactly. And that's where it gets really interesting. Because if every seller in a market is using a similar algorithm that's anchored to some arbitrary starting point, you could end up with prices that are systematically too high or too low for no good reason. The market doesn't self-correct as cleanly as classical economics would predict. Luna: So what do we do about it? As a consumer, as a regulator, as a human? Lucas: For consumers: know your numbers. Use independent sources – not the seller's suggested price. Use tools like Kelley Blue Book for cars, Zillow estimates for homes, Glassdoor for salaries. For regulators: there's a growing call for algorithmic transparency, so that if an anchor is being set, it's visible and auditable. And for all of us: just being aware that the first number matters – and that it might be arbitrary – is a huge step. Luna: It's like the old saying: 'the first one who speaks loses.' But it's more nuanced than that. The first number sets the stage. Lucas: Right. And the stage can be set by a random wheel, a phone number, or a chatbot's training data. So go into your next negotiation armed with your own anchor. The research says it's the single best way to level the playing field.