Because at maturity, the bank gets back its money. So it is perfectly valid to say "in ten years, this $100m bond is worth $100m...and I intend to hold it for ten years, so it's worth $100m [equivalent] today". The "I intend to hold it" is the relevant part of the valuation, though.
But that valuation model is not perfectly valid. It's only partially valid under limited scenarios.
As many comments have already pointed out, the issue is that you have bank customers with demand deposits. The customers can demand withdrawal of their money anytime -- without advanced notice. In other words, SVB is not a hedge fund that has the customers deposits contractually locked up for 10 years.
Therefore, the "10 year bond held to maturity" becomes invalid if the bank has to sell them prematurely at distressed discount prices -- to meet liquidity requirements of demand deposits.
You can't use value securities as "mark-to-intended-optimal-future" as an alternative to "mark-to-market" for purposes of insolvency risk calculations.
Its valid under the applicable regulations. However, it is one case where having adequate capital under those rules was not backstopped by available liquidity measures from the Fed.
One thing that seems to be generating less commentary is that, in the wake of the SVB collapse, and virtually simultaneously to the announcement of the systemic risk exception for SVB by the FDIC/Treasury/Fed, the Federal Reserve also announced a generally-available liquidity backstop program for this kind of hold-to-maturity assets.
> You can’t use value securities as “mark-to-intended-optimal-future” as an alternative to “mark-to-market” for purposes of insolvency risk calculations.
To the extent that refers to valuing the class of assets at issue at their par value, and to the extent that that was true last week, its not now.
I think this is exactly correct, but I dont think that is the purpose and scenario reported on their financial statements. I think it is fine to report valuation in terms of "mark-to-solvent future", as long as the appropriate data is provided to enable insolvency risk calculations, and the "mark-to-solvent future" model is not presented or confused with a insolvency risk model.
If an investor does not understand how HTM assets are accounted per regulation, but they are accurately reported, confusing the models is an investor error, not a bank reporting error.
My understanding is that banks provide clear reporting, and are transparent with their HTM portfolio.
HTM securities are typically reported as separate noncurrent assets; they have an amortized cost on a company's financial statements.
But your deposit demand is unpredictable and must be modeled. I don’t think even Taleb would have the scenario of “on Friday, you’ll lose $40b of deposits.” The bank would have had to be sitting on billions of T-bills, which ain’t gonna happen.
It seems like every bank is a tweet away from destruction.
Yup, and definitely anticipate there will be major new regulations in this area. A huge part of SVB's book of bonds were categorized as "Hold to Maturity". And, legally, if you mark bonds as HTM, you are not allowed to hedge against their interest rate risk. Basically, the regulations say that if you're hedging against interest rate risk, you don't really intend to hold to maturity, so you need to put them in the "Available for Sale" category.
The fact that SVB had such a huge book of bonds at paltry rates with no/minimal hedging is just awful risk management.
It already happened. The new regulation is that the Fed now has a liquidity backstop for banks holding this asset class, using cash loans with a set maximum term against the par value.
Apparently the Fed decided “if we treat it this way for capital adequacy, and we provide liquidity backstops for banks for other asset classes based on how they are valued for capital adequacy, maybe we should do the same thing here, since otherwise adequate capital can easily and suddenly become inadequate.”
This seems like an important point that I haven’t seen mentioned elsewhere. Lots of folks have been like “these people are morons they didn’t hedge their crappy bonds.” Can you write more about this?
But it's not just that they didn't hedge their interest rate risk, it's also that they assumed that their deposit base would continue to stay the same or grow. The problem is that their highly correlated deposit base of tech startups actually all needed to take their money out at the same time when they couldn't get additional funding.
Thus, it's important to understand that banks themselves make the choice of whether a bond goes into the "held to maturity" or "available for sale" bucket. I'm not a bank compliance officer so I don't know the rules about how much they're allowed to put in each bucket, but one problem was that SVB incorrectly estimated how much liquidity they would need because they didn't plan for the risk of their deposits all needing to be drawn down simultaneously.
There are liquidity tests and SVB was probably failing them (which is probably why the FDIC was paying close attention to them), but that specific test you’ve heard about is not related.
So they certainly would have known at the time that being subject to the stricter regulations would have hurt their profitability.
It’s really not, at least not mathematically. That intentions play a role is purely an artifact of regulations.
> not mathematically
Agreed. I don't think the GP was asking a mathematical question, so "relevant" meant "to explain why MTM accounting is not the only way".
Seems like that would allow an investor to see the state of the bank more clearly.
Clearly, US Treasuries carry risk that's not been accounted for.
US treasury debt is approximately the safest. The risk was that SVB might need cash before the bonds matured. The regulations encouraged SBV to do this. Now the Fed put is re-imagined, and we shuffle on while mumbling 'nobody could have imagined'.
As I was trying to point out to the parent commenter, conflating "$100m today" with "$100m at maturity" leads to clear contradictions, like saying that a bank could earn $20m on paper simply by buying bonds trading below par value. Or to put it another way – if bank A holds $100m face value of 10-year bonds yielding 4%, and bank B holds $100m face value of 10-year bonds yielding 2% (but worth, say, $80m at market price), how can you claim that those banks are on equally good footing?
Valuing liquid bonds at par value is pretty clearly a hack to reduce volatility and increase confidence in banks' balance sheets, even if some people in the comments seem to view it as a more logical way of accounting. (Although to be clear, I don't mind companies doing their own fuzzy math as long as they give investors enough information to do proper due diligence. It's similar to the non-GAAP earnings that a lot of tech companies report.)
Equal assets should never be thought to mean equal footing. The banks will show the same number for assets locked up for 10 years, but the banks will also show that they have different returns listed on their finalcial statement for the HTM assets, and different revenue from capital!
You cant and shouldnt expect to bank comparison to be easily reduced to a single measure, or for that measure to tell you something that is captured elsewhere.
It is like expecting an athlete's height to tell you something about their speed or strength.
HTM assets tell you the nominal value of assets they are holding to maturity.
It is not intended to show how much they could raise if they had to liquidate it today. It is not intended to show what that yield is for their bonds.
There are separate line items for that.
If you change the valuation of the bonds to market value, then you lose sight of the mature value of those bonds.
Replacing athlete height with athlete BMI tells you something different.
If you put cash in a a CD with a 1 year lock in, do you list it at current value or subtract a withdraw penalty.
Do you subtract early withdrawal penalties when calculating the balance in your 401k?
At the end of the day, a list of your asset values is not the same as how much you could liquidate those assets for today.
That would be a list of liquidatable assets.
HTM assets are called out separately on the balance sheet specifically to highlight that they cannot be easily liquidated.
They are still worth $100m at maturity. $100m in ten years is (usually) worth less than $100m now. Do you really want to pretend that a ten year bond you purchased when inflation and the interest rate were near zero, is worth the same when inflation and the interest rate go up to say 10%? What about inflation of 100%? In nominal terms you’ll get back your capital, but in real terms you will get back only one thousandth.
To correctly value them now you need to calculate the NPV.
You can value something at its current market value, if the asset is one that has such a thing. And fair market value will generally correspond to what you would estimate to be net present value, plus whatever risk premiums and holding costs and so on that the market is accounting for.
According to Merrian-Webster [1]:
calculate: 1 b: to reckon by exercise of practical judgment : ESTIMATE
But sure, let me clarify it to: you can’t calculate a precise NPV.
You can only estimate one.
Which, when we are trying to do things like ‘calculate the total assets a bank has’, makes the net present value of their assets a not very reliable number to use.
I wasn't taking a math test. I was saying something about NPV using a common meaning of an English word. You chose to ascribe a different meaning to that word, and pedantically - and incorrectly - tried to correct me.
$1 after inflation is still $1. It is just that the value of $1 is now different.
As long as you hold to maturity, the number of dollars does not change.
If you report your Holdings in terms of dollars, they are always accurate as long as you hold.
If someone tells you they have $100 maturing in 10 years, it is Trivial for you to do the npv calculation yourself with your speculative model of what inflation will look like over the next 10 years.
In nominal terms. In real terms you have to adjust for inflation. [1] is a starting point if you want to read more.
> As long as you hold to maturity, the number of dollars does not change.
A dollar now is not the same as a dollar 10 years from now. [2]
[1] https://en.wikipedia.org/wiki/Real_versus_nominal_value_(eco... [2] https://en.wikipedia.org/wiki/Time_preference
If you buy a 10 year bond today, you will get $100m dollars in 2023.
In 2033, that $100m will have a real value of $100m 2033 dollars on your balance sheet.
HTM assets are reported in the nominal purchase price today, which is also the real dollar value if you calculated it on the day of maturity.
I agree that if you estimated the net present value of $100m 2033 dollars, it would be worth less in terms of 2023 dollars.
This brings us back to your earlier question
>Do you really want to pretend that a ten year bond you purchased when inflation and the interest rate were near zero, is worth the same when inflation and the interest rate go up to say 10%? What about inflation of 100%? In nominal terms you’ll get back your capital, but in real terms you will get back only one thousandth.
YES! This way the asset sheet is always correct in how much you will get for the asset. If you have a $100 nominal bond it is worth $100 dollars. That is true today, and that will be true on the day of maturity. It will be true every day in between. The dollar value of the asset remains constant at the time of reporting.
Why would you want to report the net present value of that asset at maturation - which sounds like what you are suggesting?
The point of listing your bonds on your balance sheet isn't to estimate profit or returns, it is to list your current asset allocation.
You have a separate line item for revenue coming from those bonds. You have a separate model entirely for calculating ROI and profitability.
If you made a spreadsheet of your current asset allocation today, how would you list money locked in a 10 yr CD. As the dollar amount in the account, the value if you were forced to pull it out and pay a penalty, or some time shifted valuation?
you wrote 'real terms' when you meant 'nominal terms.'
> $1 after inflation is still $1. It is just that the value of $1 is now different.
That is why we distinguish between 'real value' and 'nominal value.'
>That is why we distinguish between 'real value' and 'nominal value.'
What will the real value of a $1 bond be on your balance sheet the day it matures? exactly $1
The real value in future-date dollars will be $1. In present-day dollars it is likely to be less, the ability to refer to which distinction is the purpose of the formal difference between nominal and real value.
Initial time for real dollar caluculation can be anything. You can ask what your real dollar salary is relative to 1950, or relative to 1951, or yesterday.
You can ask what was the real dollar salary in 1951 relative to 1950, or 2051 compared to 2050.
While I meant to write real dollars, I wish I wrote nominal, based on how much confusion it caused.
I still stand by the idea that it is silly, and not very useful to put a future return on investment on an asset list in to 2023 dollars using a 10 year inflation projection.
Then the reported asset would fluctuate based on your model, and you already know exactly where it will end on the maturation date.
Regardless of which day's dollars we use as a baseline for comparison, a bond issued under at a lower interest rate is discounted relative to the same bond issued later at a higher interest rate. Quibbling about how we express this valuation suggests you don't understand this difference, but this difference is important to understanding the current day banking crisis.
> While I meant to write real dollars, I wish I wrote nominal, based on how much confusion it caused.
You still seem unaware that these meanings and words are a very well understood convention that you violated. The only confusion was the confusion you had in the meanings you assigned to the words.
> I still stand by the idea that it is silly, and not very useful to put a future return on investment on an asset list in to 2023 dollars using a 10 year inflation projection.
The main thing is that these valuation rules exist for reasons and while we could debate which rules are good etc., understanding the basics of bond valuation and the common terminology we use to discuss them is a minimum prerequisite and I'm still working on getting you on board with the basic terminology every one else is using.
There isn't much discussion to be had without a common vocabulary.
I fully understand how bond market prices are impacted by interest rates. What most people seem ignorant of is the fact that bonds are not simply market trades asset, but are also have a value at maturity. Most people don't seem to know that HTM assets are reported separate from securities available for sale, which ARE tracked at market value.
And then there's the even stupider idea that the value of long-term assets should be listed as the maturation date npv, as if you can just look up future inflation rates for the next 10 to 30 years.
It seems obvious to me that if you never intend to sell a bond, the maturity value is a measure of interest.
Do you have anything else to add to the discussion, or was that your only point?
The reason this matters is because in the case of a bank who needs the funds to operate then they very much might need to sell the bonds, or revalue them at NPV because of statutory requirements.
This entire discussion is because the NPV of HTM assets is now relevant.
Most importantly, Banks already DO report the unrealized losses and Fair market value on HTM securities. Just not in the assists section, but in a dedicated section on the HTM assets. It is not some big secret.
You can even look at it in silicon valley Banks filings if you want(1). They break down the HTM losses and fair market value plain as day starting on page 125.
At the time of filing, the reported a mature value of 91 billion, fair value of 76 million, and unrealized losses of 15 billion. They break it down by the duration of maturity. Everything someone could ask for is there.
It seems to me that this whole question of reporting fair market value instead of maturity in the asset table comes from people who have never read a 10-k filing and think there is some conspiracy.
SVBs situation should have been no surprise to anyone looking. The real conspiracy is their HTM position was common knowledge.
https://www.sec.gov/Archives/edgar/data/719739/0000719739230...
The day you are paid, you will still get handed exactly $100m million.
Every day between now and then you will still have exactly $100m in bond holdings. How many cheeseburgers you can buy with that number of dollars may change from day to day, but the number of dollars will not.
(Someone else is offering 100 future for 80 current because they have a forecast about cheeseburgers, sure. But you don't have to agree with that forecast to take their deal; the deal looks even better for you if you don't agree.)
Think of how you would report a non-transferable asset with maturation on your balance sheet?
How would you report savings in a CD with a steep early exit penalty?
Herein lies the difference between a list of assets, and a list of asset liquidation value.
I understand NPV. I'd edit my post to put "$100m equivalent [NPV] today" to makes it clearer what was meant by "equivalent", but it's too late, and that's precisely what I nodded to with "equivalent" -- no need for jargon to get the sense across.
> These considerations are precisely what marking to market captures.
Of course. And they are -- in theory -- exactly what GP seemed to be asking about. Valuing an instrument at its NPV (NOT MTM) is perfectly reasonable...as a starting point. GP was questioning that. As everyone has pointed out, and anyone who's bootstrapped a yield curve or traded bonds (I have) knows, there are a ton of nuance and caveats to this, but the GP was not dealing with those and they are not relevant to GP's primary point/question.
In the extreme - obviously you can't buy a call option and say 'I intend to hold this until it's $10 in the money, so it's actually worth (time-adjusted) $10'. What's the difference, besides probabilities of outcomes?
So, can the bank claim it has $1,000 now?
To some degree this makes sense, because if the maturity timelines and classification are correct the bank is only losing opportunity cost and inflation-adjusted dollars. Not actual dollars.
Meanwhile, if they need money from the fed it is offered against collateral based on mark-to-market value.
Or that the bank should be able to say "depositor X transferred $100 million to Chase, so we sent Chase a wire for $10 million and treasuries marked HTM worth $90 million" or something.
Why? HTM bonds are not cash so they are not interchangeable. This is like if you were forced to accept a 10 year IOU in place of cash from your employer.
Banks exist at the whim of the gov't, it can require things for stability.
With your proposition, I think it would be a terrible idea and reduce stability. I don't see how letting Banks unload bad assets on each other does anything to help the situation. The Bad Assets just follow the customer wherever they go and some unlucky Bank get stuck with them when the customer converts their cash to something like stocks or buys a cheeseburger.
At best, you have just turned bonds into the same thing as cash, defeating the purpose of bonds to begin with. You buy a bond and are paid interest because you can't use it as cash.
At worst it would be chaos. Everyone buys bonds and holds them when the value is good. When the value is bad, everyone forces each other to take over their bad Investments.
What is the difference between what you are saying and just buying back the bonds before maturity?
Anyway, governments do usually have all kinds of lines of credit against bonds. And when there is a difference, it's for the benefit of the government.
Or the bank could have bought TIPS instead, I guess.
AFAIK, the thing the FDIC does is confiscate banks. Do they do anything else?
In this case, all depositors with no limit.
They don't lend money, don't put money in it, don't do anything else. They take the bank for themselves, and then proceed to pay the depositors.
Yes, the Fed basically did this.
See the recently (Sunday) announced "Bank Term Funding Program", which basically says if banks hold securities from the Federal government, the Federal Reserve will accept it as collateral for a loan, the collateral valued at par.
https://www.federalreserve.gov/newsevents/pressreleases/mone...
The Proposal is closer to the idea that you should be able to pay your cash debt with stock valued at your purchase price, and not at the current market price.
It would mean that a bank that owes another bank $100 could instead pay with Bond currently valued at $50, and then go out on the market and buy two identical new bonds with the money they saved.
The repo market is vast and generally massively liquid; trillions of dollars of funding per day.
In the olde days that was the bottom drawer where you would stuff the losing tickets at the end of the day and hope that they were in the money tomorrow.
I agree with your main point.
You'd have to be careful with regulation around this, because what you might end up with is banks preferentially seeking assets for which there is not a liquid market so they can pretend they are worth more. That's... not an obvious improvement.
The Basel accords are supposed to establish a risk-oriented way of measuring and controlling capital risk limits across asset classes. SVB and other regional banks fought heavily against being subject to this kind of oversight. I think it makes more sense to rework Basel 4 based on this failure rather than change the accounting standards.
Finance crimes are low tech and have not evolved much as they don’t need to; there’s no policing.
Have a go at it, elites scream communism and the like, and rile up the 2nd Amendment fan boys, only to throw the ones that go over the line in jail to keep up appearances.
I’ve been noting and watching this same social ebb and flow since the 80s. The kids/teens then who soaked that reality up live it still today. IMO memory is why we had a mini-Reagan in Trump grow so popular.
The reason such things you point out are allowed is they’ve always been allowed from the perspective of those benefiting from them. If the system was stable and accountable to the masses, the phony winners rich off mathematical inference but too inept to keep themselves alive would of course be subject to a terrible regime should elites be required to pull on their boot straps; a figurative identity of being coddled is all they know!
What is the alternative ? To have the banks trading on a millisecond basis so that the mark to market value of assets is positive ? What about the transaction fees ?
A bank is illiquid if its holdings require time to sell at "market price". Selling things like mortgages requires time because the buyer has to perform due diligence. If i have to sell mortgages right now, I'm going to be getting an awful price for them.
OTOH, i can sell treasuries in a fraction of a second at the ask price minus epsilon. It's not a liquidity problem, its an assets < liabilities problem, (as you clearly state, I'm just frustrated by the discussion here).
Finance is an imaginary staircase that only works until we look down and freak out. We don't document the robustness of the stairs because the stairs aren't real.
CDs are a thing. Unpopular because their rates sucked but that hasn’t always been the case.
Long-term loans become like short term loans closer to maturity. A mature bank should have a fair amount maturing every year, mortgages 24 or 23 or whatever years ago. Along with some early repayments or reissuances from people moving. Or 4 years ago for a vehicle, etc.
Many of their assets aren't really fungible either. Mortgages are the canonical example: yes they can now be turned into MBS, but your local credit union just issues and holds them). The same is true of the commercial loan to the local stationary business. As far as I know those aren't bundleable into commercial paper securities. The debtors do the same: they don't treat their loan as having any market value at all beyond what it had when issued. House prices generally don't mark to market except in places with property tax assessment.
I was on BoD of a CU. About the only thing we didn't resell were loans which did not conform. The business was to originate, quickly resell, and do some more.
- Enron used "creative" accounting and mark to something style procedures to create fake valuations
- They go out of business.
- Regulators say, "Hey! Now you need to mark to market always!"
- 2008 happens. Markets for things like CDOs and CDSs dry up almost overnight. At the very least most of the liquidity is gone and spreads get VERY big
- B/c of the above coupled with rules of "you need to mark to market" and "if value falls X% you have to sell", lots of selling happens in low liquidity environments and therefore prices fall more, the downward cycle begins
- Post 2008, people realize that "mark to market, always!" maybe isn't the best idea.
- I would imagine, that's why the Govt is saying "Ok, we will pretend that your assets are worth par value". They don't want to trigger the downward spiral.
It's also interesting to note that the Govt made money on many of the assets they bought in 2008 at well below par value. The implication is that those assets were undervalued when they bought them. Again, I would imagine this is a selling point of the idea "the par value is probably not that bad price to pay for these things now".
This is a _good_ thing. We don't want a house of cards that's so fragile as soon as it looks like it it's going to fall down, we pour glue all over it and prop it up with cardboard.
Assets need to fall in value so that the next guy can have a chance to thrive. Some bank collapsing is an opportunity for some of the younger folks to buy up a piece, or leave with a team and a book of customers.
We should have this happen regularly so that it isn't an earthquake each time.
Not getting triggered by a flash crash seems more robust, and getting triggered by it more fragile? Some of the time, anyway.
Isn’t this just a way of saying “nobody wants to pay what I want to pay me?”. Unless it’s actually worthless, there’s a buyer, you just may not like the price.
The liquidity on my used socks is gone and spreads are very BIG. Why yes, I won’t sell for less than what I paid for them new, but it’s the market that’s failed, not my insane pricing demands.
- within next 15 minutes - within a year
Each time frame will definitely result in different price. All of them will be what market was willing to pay, but prices are somehow still different.
Forced sell does have an impact.
One guess: many in the startup ecosystem have extracted liquidity from lofty valuations (via loans?), marking down those loans would cause too many other assets to decline in price (Bay Area real estate for example).
If a bank that makes a mortgage loan has to buy an interest rate swap that zeros out the interest rate risk on that loan, then the replicating portfolio is basically ... nothing ... right?
Banks have to be long duration risk to be useful.
Have a read of Matt Levine:
The “let’s do something about it” could actually be business as usual. Look at the “deposit beta” in that article: when interest rates increase, mean rates banks pay on deposits increase less. So you can have a bank with a low mark-to-market value (assume all deposits liquidate at par and securities are sold at mid-market), but that ignores the value of the enterprise itself. Or you can project forward (to a specific time or times, not necessarily to the arbitrary maturity of each security), and you may well find that you end up with enough money at every point in the future to be quite comfortable, assuming your deposits stick around and continue to earn less than market interest.
Imagine you worked at a magic trading desk where you could issue a very special bond: you borrow money right now, and you pay a floating interest rate that is set at 1/2 the federal funds rate. This is a great deal, but it comes with a catch: the bond holder can call the debt at any time, which they will do on occasion at random but will do en masse if they don’t like you. Also, people can lend you money on these terms and you have to accept the deal. How would you make money on these? How would you account for them?
I agree that HTM accounting, done carelessly, can lead to wrong conclusions.
Allowing the bank to pretend it has more assets than it actually has seems to be an invitation to hide risk. If they had to MTM their underwater bonds, they would would have been pushed to raise capital earlier, or they would have cut their losses earlier.
It should be straight up "I have these deposit liabilities, I have this book of assets, oops, my assets are down a bit, lets do something about it". Instead of "I'm gonna run the gauntlet and hope the business survives until these bonds come in".