The Banks Wall Street Bets Moment: How Contagions Spread at the Speed of Social

Tobias Fan
7 min readMar 23, 2023

--

Social media has forever changed the speed and the means by which we view and manage financial crises. Banks need a way to tactfully manage what is now being coined “social media risk”. Viral games of telephone, played at an exponential scale allow very little time for nuanced investigation or thoughtful reaction. By incorporating social media into overall risk frameworks, banks can help shape the narrative and public perception through pre-emptive and transparent engagement with their customers. Aggregation and monitoring tools like LunarCrush are becoming increasingly important to monitor for early signs of trouble and navigate this landscape of rapid information ingestion and uncontainable spread.

Recent collapses of some of the largest crypto-associated banks ($SIVB, $SI, $SBNY) have sent jitters through tech, crypto, and banking sectors. After last week’s debacle, many startups were left wondering whether they would be able to meet payroll obligations, while regional banks caught whiffs of a bank run for the first time since the subprime mortgage crisis.

Crowdsourced interpretations of information regarding $SIVB’s financial health gave depositors a crash course (no pun intended) in Game Theory 101 — putting many startups and tech companies in the shoes of a prisoner on a textbook payoff matrix: withdraw deposits now or risk holding the bag.

On Mar 8, Moody’s downgraded $SIVB bank deposit and issuer ratings 1. On the same day $SIVB announced a proposed sale of $2.25bn in stock, along with a balance sheet repositioning that showed a $1.8bn realized loss from selling fixed-income assets that had lost significant market value with the recent hike in Fed rates 2.

Like many bank runs before this one, depositors, many of them VC-funded startups and crypto companies, wasted no time playing a game of chicken. Within a day or two, $SIVB balance sheets were drying up and the stock was tumbling — by Mar 10 regulators had closed $SIVB and taken control of its deposits. A slew of regional banks followed suit, taking massive drawdowns in market value ($FRC, $WAL, $CMA, $ZION) before recovering on the news of an FDIC backstop in deposit insurance. And while there have been similar instances of bank runs in financial history, what made this one so exceptional was the speed at which it happened, and the medium on which the contagion spread.

Financial contagion spread virally this time, over social media. Social media platforms like Twitter dominate the crypto and startup space and the velocity of information spread (as well as different interpretations of it) is arguably an order of magnitude higher than traditional news or linear media. Even regulators have acknowledged the impact of social media on the most recent crisis.

House Financial Services Committee chairman Patrick Henry admitted “this was the first Twitter fueled bank run” 3.

For reference some may remember Washington Mutual, which experienced similar deposit outflows during the 2007–2008 financial crisis. Like $SIVB, WaMu held over $188bn in deposits but was beginning to write down large losses due to defaulted mortgages. When Lehman Brothers collapsed on Sept 15, 2008, WaMu depositors began withdrawing in droves — taking out over $16.7bn from checking and savings accounts (~11% of total deposits) over a period of 10 days 4. The speed of the outflows was unprecedented at the time, ultimately leading to WaMu’s bankruptcy. Contrast this with $SIVB — where depositors attempted to withdraw $42bn in a single day, equivalent to 25% of total deposits 5.

This time around, it wasn’t just the velocity of information that was unprecedented but the proximity of social distance (between Twitter followers, friends, subreddits) that helped the news spread like wildfire. Whereas before in traditional media, news was spread from centralized parties to the masses (a one-to-many transaction) — social media is a many-to-many transaction and the social distance between sources is much closer, lending these assets a level of social proof that hasn’t always been present in traditional media. And once the information reaches critical mass, it essentially becomes truth as the inertia of the spread causes the story to become ubiquitous. As they say, a person’s perception is their reality.

You can see examples of this rapid spread, even early signals of it, by monitoring aggregate social media data. Take for instance, Silicon Valley Bank:

In red we have social media volume (individual tweets/posts/news articles across Twitter, Reddit, 1000+ news sources) and in green, the price of the $SIVB stock. At a cursory glance you can see an after-the-fact spike in social media volume. However zooming in further, you can start to see growth in chatter before the parabolic spike in social media volume (signaling the powerful velocity of social media reach):

Zooming in even further into the days leading up to $SIVB’s large drawdown, you can start to see the whispers of industry insiders and analysts before a large relative spike right at the moment the stock begins to drop. By doing a bit of additional investigating around the time the social media activity started to pick up, an analyst may have been able to glean hidden gems like this one: https://twitter.com/SaaSletter/status/1633141509909524480 by Matt Harney,

or this one https://twitter.com/rusilsarkar/status/1632675515785154560 by Rusil Sarka that may have raised early red flags for any party involved with the bank.

The data here is courtesy of the LunarCrush API, a social media monitoring engine that provides access to accurate, high-latency data on over 4000+ crypto assets, 300+ NFT collections, and 700 stocks.

Having a tool that analyzes millions of real time, user generated data points across multiple social media platforms each hour, and filtering them through spam-detection and sentiment-classification machine learning engines is no small feat — and the insight potential is massive (check it out here: https://lunarcrush.com/).

Similar patterns can be found with the stocks of regional banks that were affected by the fallout of $SIVB, namely banks like First Republic, Western Alliance, Comerica, Zions, and PacWest — who experienced up to 65% drawdowns in a day. All of these large drawdowns were preceded by parabolic rises or sudden upticks in social volume, demonstrating the speed and conviction of social media contagion:

Monitoring for outsized or abnormal increases in social chatter can help catch early whiffs of quiet (but soon to be not quiet) developments. These large deviations in social activity often carry with it self-fulfilling tidbits of alpha. By capturing mass attention and spurring action in a short period of time — a game theoretic dynamic is often generated where the first person to act usually has the most to gain (or the least to lose) and makes action or participation more attractive for the subsequent party.

As the dust settles on a rough week in the banking sector — one thing is clear. Banks need to meet customers where they are and manage narratives and crises quickly in their preferred medium. A good communication strategy may have helped stem the fallout, but the key would have been to first catch these narratives early in their development. By using intelligent aggregation tools like LunarCrush and maintaining a strong, transparent social media presence, industry participants can hopefully be quicker and better prepared to react and manage these crises in the future.

Toby Fan, Web3 Strategist

@yupitstobes

Toby Fan is the Head Web3 Strategist at LunarCrush which provides aggregated real time social media data on crypto, NFT, and traditional equities. He is a graduate of UC Santa Cruz where he double majored in Econometrics and Information Systems, and helped lead departmental research on commodity market dynamics in China and the US. Toby is also an active contributor for CoinMonks (https://medium.com/@tobyornottoby) and a member of the BlockBros DAO.

Sources:

  1. https://www.moodys.com/research/Moodys-downgrades-SVB-Financial-senior-unsecured-to-Baa1-from-A3--PR_474590
  2. https://www.sec.gov/Archives/edgar/data/719739/000119312523064904/d429121dfwp.htm
  3. https://financialservices.house.gov/news/documentsingle.aspx?DocumentID=408652
  4. https://web.archive.org/web/20081001163230/http://files.ots.treas.gov/730021.pdf
  5. https://dfpi.ca.gov/wp-content/uploads/sites/337/2023/03/DFPI-Orders-Silicon-Valley-Bank-03102023.pdf?emrc=bedc09

--

--