Why did FTX go bankrupt? Analyzing the company’s downfall through Herd Behavior and Bayes Theorem
FTX is a cryptocurrency exchange that allows customers to exchange dollars for various crypto currencies such as Bitcoin. Today, FTX – one of the largest and most powerful crypto exchanges in the world– officially filed for bankruptcy. In addition, CEO Sam Bankman-Fried resigned from his position. Due to a run on deposits, the company valued at $32 billion suddenly collapsed. Here, we can take a step back and analyze the intersection of bank runs and herding behavior.
Herding behavior – also known as “following the crowd” – occurs when people are influenced by each other’s behaviors and decisions. People are unquestionably influenced by the decisions of others, particularly those who are in the same network, since they have ties to each other in some way or another. Suppose several individuals began withdrawing their crypto holdings from FTX at once. There can be a host of reasons why they have chosen to withdraw their holdings, but people in the network that have not withdrawn their money now are worried that there might not be enough funds to cover their own accounts. Here, an information cascade begins, because the next people in the network begin to follow the behavior of the individuals they’re connected to, who have decided to withdraw. At some point, this herding behavior becomes so significant that a gap is created between the FTX funds available and the demand for withdrawals. FTX could no longer meet these demands, so a shortage of currency occurred, and they have now gone under.
For traditional banks, the federal government instituted FDIC insurance in the 1930s as a result of the Great Depression. This was created as a way to prevent runs on banks, where they have increased the incentive to follow the crowd and not withdraw. Crypto banks like FTX are subject to herding behavior because they are not backed up by FDIC insurance.
We can also apply Bayes Theorem to the FTX meltdown by building a mathematical model for how this information cascade occurred. Bayes Theorem tells the probability of an event occurring (Event A) given we know a different event occurred (Event B). For FTX, Event A would be the outcome that FTX goes bankrupt. Event B is the amount of withdrawals from FTX. This value would change depending on when it is measured. Once the information cascade begins, and everyone starts withdrawing from FTX, Pr(B) increases, so Bayes Theorem tells us that Pr(A | B) increases. This morning, the day that FTX officially went under, the dramatic amount of new withdrawals suggested that the probability of FTX going bankrupt was a near certainty.
Although we don’t have concrete values to determine exactly what factors caused FTX to implode, using the tools of herding and Bayes Theorem provides an interesting perspective as to why the once valued $32 billion crypto exchange is now officially bankrupt.
Source: https://www.nytimes.com/2022/11/11/business/ftx-bankruptcy.html