Live data from Hacker News

Bank Failures Visualized

observablehq.com

21–30 of 424 posts

Re: Bank Failures Visualized

#21
post #11

That needs to go back to the 80s to capture the SnL crisis. It dwarfs 08 in bank failures. It better indicates the conglomeration of the many banks into the few we have today.

Yeah these sorts of figures always frustrate me because the historical context for these things is so much broader, and it seems obvious to me to go back to the 80s, if not earlier. There's a better figure here I think: https://www.pewresearch.org/short-reads/2023/04/11/most-u-s-... Also some nice figures here: https://www.bankingstrategist.com/history-of-us-bank-failure...

TIL there was a different "First Republic Bank" that failed in 1988. This name carries a curse now.

Re: Bank Failures Visualized

#23

That needs to go back to the 80s to capture the SnL crisis. It dwarfs 08 in bank failures. It better indicates the conglomeration of the many banks into the few we have today.

> conglomeration of the many banks into the few we have today

There are still thousands of banks in the USA. I dont really see why there should be more than 100. Canada has 5 big ones and a few dozen tiny ones. Same in UK and Australia.

Re: Bank Failures Visualized

#25

That needs to go back to the 80s to capture the SnL crisis. It dwarfs 08 in bank failures. It better indicates the conglomeration of the many banks into the few we have today.

08 was artificially low because many banks got merged at a fire sale. Wachovia, Merrill Lynch, Bear Stearns, and National City stick out. Other financial institutions got essentially nationalized and stock became mostly worthless like Citi and AIG, although the government sold most of their stock in 2011 Credit Suisse is about the same size as SVB, Signature Bank, and First Republic combined but it got “acquired” by…

Credit Suisse is not included because it is a Swiss bank, not an FDIC insured US bank.

Lehman Brothers is also not included because, even though it was a US bank, it was an investment bank with no FDIC insured deposits. It was around the size of all of this year's failures, combined.

As you note, bank bailouts that were not FDIC bankruptcies are also not included.

Re: Bank Failures Visualized

#26
post #17
post #4

r should be sqrt(assets). It is.

It is! That's the default radius type for the dot mark on Observable Plot. To play with it, add `type: "linear"` to the dict on line 6.

Thanks. I guess rename 'r' to 'area' will make it easier to read.

Re: Bank Failures Visualized

#27
post #4

r should be sqrt(assets). It is.

I generally love Mike Bostock's work, and this is a beautiful visualization, but I think this comment points out why circles are just a bad way to represent anything visual other than pizzas: it's a leaky abstraction (which is a real sin in visualization)

You have to know how the chart is generated in order to correctly interpret what you're seeing. The value can be the radius, diameter (how pizzas are measured) or the area itself! All of these choices lead to different interpretations of what you're seeing.

Then, on top of that, we're just not great and visually comparing areas. Pizzas are a great example of this. A 12" pie and a 16" pie don't look that different but you're talking about nearly twice the amount of pizza! A good example in this visualization is Signature vs SVB

Then on top of that we have the dimensions of the circle forcing the visual to overlap on the x-axis without this meaning anything concrete.

All that you can meaningfully take away from this visually is "2008 had one big and a ton of small failures, and currently we're seeing 3 big, but not quite as big failures", and unfortunately there's not a lot more you can get out of this. Because of the x-axis problem it's hard to even tell if WaMu preceded some of the others or not.

Re: Bank Failures Visualized

#28
Community-scale banks (aka credit unions) are a better idea for local-regional communities (aka cities and towns and agricultural regions) because their managers have to live with their clients.

Take a community of 100,000 families, in an economic system where they're all collecting income and paying bills and so on. The idea behind a bank is that they hold the community's money securely while making their own money by providing loans at a greater interest rate than the interest rate they pay out to their depositors. Fundamentally, this means the bank is dependent on the fact that customers don't try to withdraw all their money at once (as the above dynamic means the bank is never able to meet this demand).

FDIC guarantees by governments are intended to bolster this customer trust, as if everything goes sour the deposits (to a certain limit) are guaranteed. The important point is that the regional bank has a profit model based on giving successful loans to individuals and small businesses for things like houses, cars, business expansion, etc. The bank will of course want to know a fair amount about the people they are loaning to (drug addicts are not good bets, etc.) in order to ensure they get paid back.

I hear people laughing in the background about the naivete of this picture in the modern American financial system. Since this post is already a bit long, go watch the movies "Margin Call" and "The Big Short" to get an idea of what's actually been going on.

Re: Bank Failures Visualized

#29
Some people want smaller banks because "too big to fail". While those same people are already freaking out when two banks fail.

I'm taking the downvotes: just open an account at JPMORGAN. Banking is serious business your artisanal cornershop credit union can't hack it.

Re: Bank Failures Visualized

#30

That needs to go back to the 80s to capture the SnL crisis. It dwarfs 08 in bank failures. It better indicates the conglomeration of the many banks into the few we have today.

Round 2 of the 2023 bank crises might be commercial real estate loans like the 1980s crises. Some downtowns are still 40% vacant. This more a consequence of covid than bad bank behavior.
Post reply on HN