Why We Herd
After Sam Bankman-Fried, the idolized founder of cryptocurrency exchange FTX, was charged with fraud, a CBC podcast set out to investigate “How did Sam Bankman-Fried happen?” The final episode, “It Takes a Village”, opened with a complaint from FTX investor, Christ Keuchkerian:
I was angry not only at SBF, as they call him, but at all these people that convinced millions of customers to put their money in there…You’re talking about the [Ontario Teachers’] Pension Fund. I’m talking about Sequoia. I’m talking about CNBC. I’m talking about Kevin O’Leary. These are a lot of people. These are a lot of institutions. These are a lot of companies…
I only started committing when it looked like there was people that I look up to, in the sense of investments…I feel tricked by FTX and everybody that I trusted...actual people who do investing as a living.
The spectacular collapse of FTX swept the crypto industry, investors and customers in its wake. The write-offs were eye-watering: Singapore sovereign fund Temasek Holdings – $275 million, venerable VC firm Sequoia Capital – $210 million, the Ontario Teachers’ Pension Plan – $95 million, technology investor Tiger Global – $38 million, and celebrity Kevin O’Leary – $18 million.
We often attribute bad investments to duplicity and hubris: investors, blinded by unrealistic gains, allowed themselves to be seduced by charlatans. The truth, however, is more benign. Hype works because of a social phenomenon known as information cascades.
What are Information Cascades?
Information cascades occur when we make decisions based on earlier actions of other people. We form beliefs based on information privately available to us, as well as our observation of what others do. We are influenced by others’ behaviors in countless situations: the shows we watch, the stuff we buy, the opinions we hold and the people we trust. Armed with these beliefs, we choose our action.
In an information cascade, the first individual in the chain chooses based on their own information as there is no prior action to observe. However, when enough people decide in a specific direction, the next individuals in sequence will succumb to the choice of the majority, even if the information at hand is contradictory. Once a cascade begins, decision makers ignore their private information and go for the exact same action as their predecessors.
Here, economics professor Ashley Hodgson explains the mechanics of information cascades with a simple game:
Information cascades are prevalent in tight circles where everyone knows one another and relies on each other for information, such as in venture capital. This was true in the case of FTX, as O’Leary confirmed in the podcast:
In the deal community and venture, we rely on each other. We don't want to do the same thing over and over again. So, we made the assumption that the due diligence done by Bain [and Company] for Tiger [Global] was good…I would say [that this is] true for all of us.
Essentially, the decision made by these seasoned investors conveyed way less information about the exchange than we might think. The millions of people who followed suit are left holding the bag.
Ignoring Private Information
A layperson with limited access to data would assume that professionals, such as Sequoia Capital, make better decisions because they have exclusive knowledge. But the intelligentsia too fall prey to herd behavior, like the case of Theranos. Founded by Elizabeth Holmes, who is in prison for defrauding investors, Theranos was a healthcare start-up. The crumbling of this $10-billion house of cards left many wondering how our smartest and brightest were fooled.
Extensive reporting revealed that skeptics were unheeded by Holmes’ distinguished supporters. Stanford University professor and founding board member of Theranos, Channing Robertson, was warned “twice” over the course of 15 years by colleague Phyllis Gardner. Prior to the founding of Theranos, Gardner had dismissed Holmes’ fantastical pitch as impracticable. Another ardent Holmes supporter was Theranos board member and esteemed statesman George Shultz. His grandson, Tyler, witnessed the dysfunction firsthand as a Theranos employee. But Shultz pressured him to keep quiet. Robertson and Shultz lent tremendous credibility to Holmes and her company, attracting millions of dollars in funding, until the fraud unceremoniously spilled into the open.
Breaking a Cascade
Decision makers who choose to follow their predecessors keep their private information hidden. This loss of information is why subsequent individuals, drawing rational inferences from what everyone else is doing, could end up at the wrong end of a cascade. However, the opacity is also why information cascades are easy to stop. A long-lived cascade can be overturned if additional information is disclosed (for instance, when someone divulges not only their action but also their private information), or if enough people deviate from herd behavior and only follow their private information.
State Space Model of Hodgson’s Marble Bag Game
The tides turned against Theranos and FTX when privileged information started making its way into public view. For Theranos, a tip to journalist John Carreyrou set off a series of disclosures, including from Tyler Shultz, that its technology was fake. The company unraveled over 2 years as lies and secrets came to light. In the case of FTX, a leaked balance sheet exposed SBF’s imprudent use of customers’ money to prop up his empire and buy political influence. Realizing this, investors and customers raced to the exit, withdrawing $6 billion within 72 hours. 8 days later, FTX was bankrupt.
The Wisdom of Crowds
In decision-making, we should be careful when drawing conclusions about the best course of action from the behavior of a crowd. This is pertinent in corporations, where many decisions – planning, spending, hiring – are made in committees where people sit around a table and discuss solutions. In these situations, it is common to go around the room asking for opinions and votes. A rational individual weighing in later may assume that those speaking before have better insights, even if they have contradictory evidence. Information cascades show that, even without the pressure to conform, we could still end up in an echo chamber, detached from reality.
As leaders, we need to find the balance between allowing experts to form their own conclusions and collaborating to reach a consensus. One way is for people to declare what they think without hearing what others’ thoughts are, before engaging in deliberation and consolidation of everyone’s independent opinions. As individuals, it is always worth asking if we are taking the same action because we possess information saying that it is the right thing to do, or because everyone else is doing it.
Further Reading:
Chapter 16 Information Cascades. Easley, D., & Kleinberg, J. (2010). Networks, Crowds, and Markets: Reasoning About A Highly Connected World. Cambridge University Press.