Comment by amluto | Hacker News Reader

Comment by amluto | Hacker News Reader

Parent Full thread

> The bank was capitalized fine (albeit with underperforming treasury assets given interest rate increases)

That’s an “albeit” you could drive the second largest bank failure ever though. If I have $100k of net assets and I lose $10k due to interest rate increases, I’m fine. If I have $100k but I lose $150k due to interest rate increases, then I’m only fine if I have some other source of income before whatever debt I have comes due.

SVB was only capitalized fine in a universe in which they could maintain a large non-interest-bearing deposit volume, and make adequate profit on it, for long enough to erase the hole in their balance sheet before their bonds matured and they would inevitably be forced to realize their loss. Or if their implicit gamble that rates would go back down would pay off. (Of course, the latter also reduces the income from said non-interest-bearing deposits.)

I don’t know how bank stress tests work, but I sure hope they would notice that the bank’s liabilities exceeded their assets even assuming said assets could be sold calmly and at favorable prices. And no, “I’ll hold them to maturity so they’re worth 20% more than they are actually worth” should not be part of that calculation.

> I don’t know how bank stress tests work, but...

SVB would have passed stress tests because they did what all banks were supposed to do. They invested in "safe" long term treasuries. The only economic environment hypothetical they would fail is, "what happens if there is a bank run and you are forced to realize losses". But what bank can survive a bank run? The Dodd-Frank stress tests don't try to ensure banks can survive bank runs, do they?

When banks invest in the long run, your assets can depreciate in value temporarily to get through short term business cycles. By investing long, a bank's balance sheet will be the average of both assets that aged poorly and assets that aged well. The average should reflect the long term growth of the economy.

The treasuries purchased in the last 2 years depreciated a lot in value and most of SVB's growth happened in the last two years, so there current balances are skewed heavily towards assets purchased right before the interest rate hikes began. If SVB were able to continue operating for say another 10 years, they would be fine even if interest rates never returned to near 0 levels. Additional treasuries would be purchased and the fraction of the balance sheet that consisted of 2020-2022 treasuries would shrink. The rest of the portfolio would probably increase in value. Treasuries purchased today are cheap and have a large upside as rates start coming down, even if rates don't fall all the way to 0. The gains from cheap treasuries purchased towards the end of 2022 and after have potential to dramatically grow in value.

There really isn't any evidence that SVB behaved irresponsibly nor that stricter regulation under Dodd-Frank would have made a difference. If SVB had been inclined to invest in riskier types of investments, Dodd-Frank would have pressured SVB to buy treasuries.

> But what bank can survive a bank run?

Not all banks are equally likely to have a bank run.

China for example restricts banks from being over exposed to a single sector, so the rapid collapse of an industry does not trigger liquidity issues as badly ad those that hit SVB.

Not dodd-frank or Basel 3 required this tho.

> SVB would have passed stress tests because they did what all banks were supposed to do

No, they wouldn’t. SVB’s duration would have triggered noncompliance with Fed stress tests and Basel III.

If SVB were solvent, they could have borrowed at the Fed’s discount window.

On paper SVB might well have been solvent until they got withdrawals. This is possible because you can account for bonds in two ways: mark to market, or held to maturity.

In the second case you don't look at market prices. This actually makes some sense. If you intend to hold your bonds to maturity, any unrealized losses will naturally resolve themselves as you reach maturity. Hence a balance sheet that doesn't mark the losses is a more accurate picture. But only IF you never need those bonds for liquidity.

Note that liquidity crunches can ruin balance sheets even without this bond-specific feature. For example if you own 10% of a company stock, and need to dump all of it, the price will drop by a lot. They were accounted using Mark to Market, but even that was too optimistic in the face of liquidity crunches.

> Hence a balance sheet that doesn't mark the losses is a more accurate picture. But only IF you never need those bonds for liquidity.

I grade this as a fail of a hypothetical CFO interview I’m running. Sorry.

> Note that liquidity crunches can ruin balance sheets even without this bond-specific feature. For example if you own 10% of a company stock, and need to dump all of it, the price will drop by a lot.

Of course, but that’s an unrelated effect.

I can lose money by buying something that loses value. I can also lose money by buying something that is hard to sell without paying large spread or causing a large market impact because I sold it. These are entirely orthogonal — I could buy securities that go up and lose some or all of the gain due to a forced sale.

SVB’s bonds went down, in an NPV sense or a mid-market sense or any other sense that considers what they’re actually worth today.

> Hence a balance sheet that doesn't mark the losses is a more accurate picture.

A dollar in ten years isn't equivalent to a dollar today, so I don't think it's really an accurate picture to claim otherwise.

You could use the held to maturity value as part of a projection of your balance sheet into the future. Although, mark to market to get current value and assume prevailing interest rates to get to future value should get you to the same place.

I agree that if 20% of your deposits are demanded in the same day, that's going to cause at least big problems if not failures to most banks.

Balance sheets mix maturity very often. Outstanding loans to others, unpaid invoices, issued bonds, loans with banks. All of these are on the balance sheet, none of them represent dollars now.

A balance sheet does not show "what is left if we needed to liquidate now". It shows some approximation of "what do we own, and what do we owe". If you fully intend to hold something to maturity (either bonds you own, loans you made, bonds you issued, or borrowed money) then it doesn't make sense to value that at market prices when you know that the price will converge to 100cents on the dollar before you sell it.

If you borrowed at 1% interest and interest rates rise, you don't mark down your debt in value. Even though the 'market value' of that loan definitely dropped.

Reply on news.ycombinator.com