Silvergate Bank to begin voluntary liquidation

Some are probably upset with it because they want Silvergate's liqudiation to be caused by crypto, and not super-safe US Treasurys. Turns out Silvergate's bitcoin loans are just fine, and it was the Treasurys that fucked them up.

Others are probably upset with it because Silvergate had the option to hold cash as cash, or cash as one-month T-bills, and if they had done that, they wouldn't have had to liquidate.

But instead they locked it up in much longer dated bonds to grab a little extra yield, and got burned when interest rates increased.

If a bank classifies their long-duration Treasurys as held-to-maturity, they don’t need to mark the fluctuating value of the bond principal to market. [0]

Also, most other banks aren’t overly concentrated in one area that could collapse and force a bunch of customer withdrawals like Silvergate was (in cryptocurrency)

If there was a problem in banking bond holdings, it likely would’ve surfaced by now with the massive interest rate changes.

[0] https://viewpoint.pwc.com/dt/us/en/pwc/accounting_guides/loa...

I should have phrased it differently. The potentially large unrealized capital losses on UST sitting on bank balance sheets has been under-reported. Banks are both required to hold UST and not always required to mark-to-market. This interest rate move has been massive relatively speaking, leaving some number of banks in a terrible capital position, just waiting for a disturbance (like the FTX implosion) to trigger run-like conditions.

This story sketches a problem that hopefully doesn't start to gather steam:

https://www.axios.com/2023/03/04/why-some-banks-are-heading-...

That's not because voice recognition doesnt work. It's been working quite well for 15 years now. Just ask the Deaf community.

The issue is even if the error rate is 1 in 100 and you scale that up to a billion users supporting thousands of use cases, you then need to maintain a large army of people to manually handle those errors.

Its a huge cost to even the largest firms. Very similar to what Facebook/Twitter/Youtube end up doing to handle content moderation. The faster systems scale the faster you get swamped with bugs you don't have bandwidth to fix. The consequences begin to multiply. Your basic Jurassic Park story plays out.

I’m not sure what I think about this. It’s literally the first time I hear about Silvergate, I don’t have strong feelings about its continued existence or lack thereof, and my vision of the ideal situation in cryptocurrency is rooted in, like, 2010, but I’m reading this and...

> It’s just last week that US regulators warned banks about this [...]. It’s almost like they knew this was coming. The regulators did not quite say “therefore, don’t bank crypto exchanges”; in fact, they said the opposite [...]. But you got the idea.

If you stay silent most of the time, but say “we might be on a collision course with an iceberg” an hour before said iceberg comes along, your advice is very valuable (it might not be legibly valuable if it’s the first time you spoke up, but that’s not the concern here). If you say a lot of things all the time and then warn about the iceberg thing a minute before it happens, well, your advice is still better than nothing, but it’s not exactly an example of amazing foresight. And surely appending “captains have broad discretion over the course of their ship, as permitted by law or regulation” to the warning should discount its value further, at least a bit.

> A run on the bank happens in, like, It’s a Wonderful Life, but in the real world of big US banks, that particular dynamic [...] would be strange. [...] The story today is that Silvergate’s customers are withdrawing their money because they are worried about Silvergate [...]. But that's not why they were withdrawing their money in late 2022, when the trouble started. Then, they were withdrawing their money because crypto had collapsed [...]. The customers — crypto exchanges — were the problem, not Silvergate. [...]. I suppose this counts as contagion from the crypto crash to the real financial system [...] It is a narrow sort of contagion: That bank is pretty much the Bank of Crypto [...]. But it is certainly the sort of contagion that regulators will want to discourage [...].

Are the crypto exchanges the problem, though? This doesn’t read like a “cryptocurrency bad” story, it reads like a “being the Bank of Thing is bad” story. Cryptocurrency exchanges are a particularly bad value of Thing here, sure, but generally speaking, if you are the Bank of Thing and an overwhelming majority of your depositors are in the Thing business, then aren’t you essentially betting that Thing will not experience short-term volatility? Most of the time, the most relevant Thing is retail deposits, and as the article mentions, there’s a whole bunch of stuff instituted to smooth that out, but doesn’t this still apply for other, deposit-insurance-disadvantaged Things just as well?

In that view, it seems like, first, there should be indeed some opinions that, if you a running a bank, being an undiversified Bank of Thing on the depositor side is not a good idea, but also, second, that if you are a bank regulator and you regulate a Thing to a point where most but not all banks will refuse Thing businesses, you are very much accepting that those few remaining banks will be vulnerable.