If the FED had to mark to market its portfolio than the government would know who to bail out. Instead they are allowed an accounting practice that is simply ridiculous. Not only, by raising rates recklessly they are doing a mistake in an attempt to correct a previous mistake, I.e. their inaction in 2021. So bottom line much of the causes are related to out of control fiscal and monetary policies conducted in Washington. Management of banks is definitively adding to the problem.
Valuing securities at historical cost vs fair value is a question of GAAP rules/standards. Couldn't regulators require banks to report securities at fair value in their call reports or TFRs or whatever they call them these days? The banks most likely know what the market value is of the securities they hold. I don't think it's a huge burden for them to write it down and include it in a big report they have to fill out, anyway. Seems like if regulators are "blinded" it's because they prefer it that way.
I thought Matt Levine's Bloomberg column on this was very useful. He makes the point that part of a typical bank's stability comes from diversity in its customer base, so that you have a good mix of people who want deposits and people who want loans, and so that desire to withdraw money is not too strongly correlated across customers.
SVB didn't have this diversity and it's a huge part of what killed them. They focused on serving cash-rich startups that didn't need loans, so they invested in bonds instead of lending, which gave them one side of the interest rate risk. And then on the other side, when rates rose, all their customers were similarly squeezed and wanted to withdraw at the same time, initially not even due to a bank run mentality but to their lack of diversity.
It's funny how lack of attention to correlations caused this failure, just as lack of attention to correlations in the underlying assets for mortgage backed securities caused problems in 2008. If I were a principles based regulator, I would try to operationalize the principle that in your risk management, you have to state and check all your assumptions about different risk events being independent of each other
Thanks for the reference to S&Ls. I worked with a couple of former FSLIC regulators in the 1990s in the xUSSR. There, there was no market to which to mark! Also there, there was no deposit insurance, though a large bank failure almost certainly would have provoked a government bailout (but more likely of bank owners rather than depositors.)
It seems like forcing mark-to-market accounting for banks would greatly exacerbate the mortality of banks, which is what we're trying to avoid. Because it will create large and arbitrary swings in its capital position that are transparent to everyone. This would clearly lead to people jumping to the "bank is in trouble" conclusion which becomes a self-fulfilling prophecy.
Seems like a better way would just be to increase capital requirements based on a general rule (e.g. interest rates and market values). It's a given that banks will toe whatever line you draw, so draw one that hits the sweet spot between allowing actual risk and causing unnecessary panics.
Indeed, 50 years from now the Fed will look back at QE plus supporting issuing record amounts of long-dated bonds at 3.5% to support reckless executive and congressional actions with the same disdain we look at the 1930s Fed actions.
What SVB shows is that at most banks and insurance cos the level of "sophisticated" allocation of safe deposits is not much beyond -- get money, invest in govt bonds down the curve at higher yield, and sleep soundly. Doubt SVB had anyone even barely able to model let alone evaluate duration risk.
This stupidity of the govt/Fed on managing the long dated side of our debt is just about equal to their stupidity of carrying another 8 trillion of debt at less that 1 year duration and betting on team transitory while simultaneously stepping on the fiscal gas pedal thereby ensuring losing any benefit to the government of not immediately having to refinance to offset pain inflicted by MTM on the public.
This is not over by a long shot.
A lot of investors opting for safe government assets have taken losses associated with the equity market without the gains. And those opting for super safe short dated government bill money market funds have lost against inflation and if there is G** forbid a default, the runs on these funds will make this look like a bagatelle -- again, those who went for "safe" govt assets will face most of the downside.
The government via the Fed incentivized its citizenry to come along at low interest rates, and is now abandoning them by using monetary rather than fiscal policy to try to solve its problems.
Seems this administration is making a habit of asking for allies in both wars and finance and abandoning them at the worst moment in the worst way.
The COVID money drop in conjunction with "zero" rates created a financial dislocation that cannot be fixed. As others have commented, millions of people have now borrowed at very low rates for very long term, and they have no incentive to close those loans early. Someone must carry those loans on their books. Who?
Hi, Arnold. I think that you've said in the past that banks perform the liquidity transformation function of taking short-term deposits and lending long term. Do you think that, nowadays, we "need" banks for this purpose and that such need justifies FDIC, capital requirements, and other regulations and subsidies? It seems like we have plenty of investment funds now to finance long term debt, whether it's bonds or securitized loans and mortgages. We certainly don't need an SVB to buy long-term bonds (not issuing loans) with short-term deposits.
What would be the (economic) feasibility of so-called narrow banks, which invest short-term deposits only in liquid money market instruments like T-Bills, combined with relying on investment funds to finance long-term borrowing via bonds and securitized loans? In such a scenario, we wouldn't need FDIC to avoid bank runs, but what would be the main drawbacks, if any, compared to the present system? Politically, the Fed quashed narrow banks recently, I think mainly to preserve traditional banking. (The Fed worried that narrow banks would drain too many deposits from traditional banks, which would then not be able to issue as many loans.)
Since the real interest rate is still negative for low risk bonds and we probably can't kill inflation until real interest rates become positive, the accounting/regulatory system will allow banks to effectively pay their customers a more negative interest rate than the negative bond rate and obtain a positive return. If the payout times for the customers become less than the bonds, the system crashes.
A bank run is just a rapid change time line for withdraws relative to the time line for bond payments. Mark to market is a measure of that potential time instability. If the inflation rate goes to zero and the value of money becomes constant, the higher interest rate long bonds will always be worth more than the deposits and the system may not go unstable and have a bank run.
It may be the highly variable inflation rates that create the instabilities as you don't know how fast the dollar may lose value over time. This is especially true when inflations rates and their variances are much larger than real interest rates.
Great post and book review here. Other commentators mislead people by speaking about banks like this as if they are normal market participants that are obliged to obey all the rules that normal individuals do.
From your review of the Ferguson book:
>Much of the capital that did not go to finance housing went instead to finance large government deficits, especially during the pandemic years of 2020-2022. Some investment also went into other financial assets. In the decade between the financial crisis and the pandemic, there was a surge in stock prices, bond prices, valuations of firms funded by venture capital, and cryptocurrencies.
So, what's next for the printers-in-chief? Probably not house construction. With the way things are going, it looks more like that there will be attempts to build up arms production capacity, but that suffers from a whole alternate universe of regulatory overhead and other issues besides.
The emphasis on ignoring mark-to-market values is spot-on in this case.
I think criticizing risk weighting is wrong. SVB didn't buy treasuries and mortgages because the government said they were riskless, they bought them because they didn't want to bother with real bank lending and those are deep and liquid markets.
This failure seems to raise the question of whether or not all deposits should be insured though. They didn't fail because of mark-to-market losses - they failed because of those plus the fact that effectively all their deposits were uninsured. We already know how to fix this if we want to.
Is this a good place to ask stupid question? I am new here, I know nothing about these issues, hoping to find some help...
I understand the first step-- SVB buys government bond/mortgages because they require a smaller capital cushion. In the absence of "mark to market" the capital cushion is smaller than it should be. Right? But I don't understand the precise steps that take us from
"Roughly speaking, if the bank paid $100 to buy a long-term bond with an interest rate of 2-1/2%, and now the interest rate on a comparable bond is 5%, the bank’s bond is worth about $50."
My first thought was "no way." After Googling I find that a quick approximation is the change in interest rate x years duration.
If the FED had to mark to market its portfolio than the government would know who to bail out. Instead they are allowed an accounting practice that is simply ridiculous. Not only, by raising rates recklessly they are doing a mistake in an attempt to correct a previous mistake, I.e. their inaction in 2021. So bottom line much of the causes are related to out of control fiscal and monetary policies conducted in Washington. Management of banks is definitively adding to the problem.
Valuing securities at historical cost vs fair value is a question of GAAP rules/standards. Couldn't regulators require banks to report securities at fair value in their call reports or TFRs or whatever they call them these days? The banks most likely know what the market value is of the securities they hold. I don't think it's a huge burden for them to write it down and include it in a big report they have to fill out, anyway. Seems like if regulators are "blinded" it's because they prefer it that way.
I thought Matt Levine's Bloomberg column on this was very useful. He makes the point that part of a typical bank's stability comes from diversity in its customer base, so that you have a good mix of people who want deposits and people who want loans, and so that desire to withdraw money is not too strongly correlated across customers.
SVB didn't have this diversity and it's a huge part of what killed them. They focused on serving cash-rich startups that didn't need loans, so they invested in bonds instead of lending, which gave them one side of the interest rate risk. And then on the other side, when rates rose, all their customers were similarly squeezed and wanted to withdraw at the same time, initially not even due to a bank run mentality but to their lack of diversity.
It's funny how lack of attention to correlations caused this failure, just as lack of attention to correlations in the underlying assets for mortgage backed securities caused problems in 2008. If I were a principles based regulator, I would try to operationalize the principle that in your risk management, you have to state and check all your assumptions about different risk events being independent of each other
Thanks for the reference to S&Ls. I worked with a couple of former FSLIC regulators in the 1990s in the xUSSR. There, there was no market to which to mark! Also there, there was no deposit insurance, though a large bank failure almost certainly would have provoked a government bailout (but more likely of bank owners rather than depositors.)
It seems like forcing mark-to-market accounting for banks would greatly exacerbate the mortality of banks, which is what we're trying to avoid. Because it will create large and arbitrary swings in its capital position that are transparent to everyone. This would clearly lead to people jumping to the "bank is in trouble" conclusion which becomes a self-fulfilling prophecy.
Seems like a better way would just be to increase capital requirements based on a general rule (e.g. interest rates and market values). It's a given that banks will toe whatever line you draw, so draw one that hits the sweet spot between allowing actual risk and causing unnecessary panics.
Indeed, 50 years from now the Fed will look back at QE plus supporting issuing record amounts of long-dated bonds at 3.5% to support reckless executive and congressional actions with the same disdain we look at the 1930s Fed actions.
What SVB shows is that at most banks and insurance cos the level of "sophisticated" allocation of safe deposits is not much beyond -- get money, invest in govt bonds down the curve at higher yield, and sleep soundly. Doubt SVB had anyone even barely able to model let alone evaluate duration risk.
This stupidity of the govt/Fed on managing the long dated side of our debt is just about equal to their stupidity of carrying another 8 trillion of debt at less that 1 year duration and betting on team transitory while simultaneously stepping on the fiscal gas pedal thereby ensuring losing any benefit to the government of not immediately having to refinance to offset pain inflicted by MTM on the public.
This is not over by a long shot.
A lot of investors opting for safe government assets have taken losses associated with the equity market without the gains. And those opting for super safe short dated government bill money market funds have lost against inflation and if there is G** forbid a default, the runs on these funds will make this look like a bagatelle -- again, those who went for "safe" govt assets will face most of the downside.
The government via the Fed incentivized its citizenry to come along at low interest rates, and is now abandoning them by using monetary rather than fiscal policy to try to solve its problems.
Seems this administration is making a habit of asking for allies in both wars and finance and abandoning them at the worst moment in the worst way.
The COVID money drop in conjunction with "zero" rates created a financial dislocation that cannot be fixed. As others have commented, millions of people have now borrowed at very low rates for very long term, and they have no incentive to close those loans early. Someone must carry those loans on their books. Who?
Hi, Arnold. I think that you've said in the past that banks perform the liquidity transformation function of taking short-term deposits and lending long term. Do you think that, nowadays, we "need" banks for this purpose and that such need justifies FDIC, capital requirements, and other regulations and subsidies? It seems like we have plenty of investment funds now to finance long term debt, whether it's bonds or securitized loans and mortgages. We certainly don't need an SVB to buy long-term bonds (not issuing loans) with short-term deposits.
What would be the (economic) feasibility of so-called narrow banks, which invest short-term deposits only in liquid money market instruments like T-Bills, combined with relying on investment funds to finance long-term borrowing via bonds and securitized loans? In such a scenario, we wouldn't need FDIC to avoid bank runs, but what would be the main drawbacks, if any, compared to the present system? Politically, the Fed quashed narrow banks recently, I think mainly to preserve traditional banking. (The Fed worried that narrow banks would drain too many deposits from traditional banks, which would then not be able to issue as many loans.)
Since the real interest rate is still negative for low risk bonds and we probably can't kill inflation until real interest rates become positive, the accounting/regulatory system will allow banks to effectively pay their customers a more negative interest rate than the negative bond rate and obtain a positive return. If the payout times for the customers become less than the bonds, the system crashes.
A bank run is just a rapid change time line for withdraws relative to the time line for bond payments. Mark to market is a measure of that potential time instability. If the inflation rate goes to zero and the value of money becomes constant, the higher interest rate long bonds will always be worth more than the deposits and the system may not go unstable and have a bank run.
It may be the highly variable inflation rates that create the instabilities as you don't know how fast the dollar may lose value over time. This is especially true when inflations rates and their variances are much larger than real interest rates.
Great post and book review here. Other commentators mislead people by speaking about banks like this as if they are normal market participants that are obliged to obey all the rules that normal individuals do.
From your review of the Ferguson book:
>Much of the capital that did not go to finance housing went instead to finance large government deficits, especially during the pandemic years of 2020-2022. Some investment also went into other financial assets. In the decade between the financial crisis and the pandemic, there was a surge in stock prices, bond prices, valuations of firms funded by venture capital, and cryptocurrencies.
So, what's next for the printers-in-chief? Probably not house construction. With the way things are going, it looks more like that there will be attempts to build up arms production capacity, but that suffers from a whole alternate universe of regulatory overhead and other issues besides.
The emphasis on ignoring mark-to-market values is spot-on in this case.
I think criticizing risk weighting is wrong. SVB didn't buy treasuries and mortgages because the government said they were riskless, they bought them because they didn't want to bother with real bank lending and those are deep and liquid markets.
This failure seems to raise the question of whether or not all deposits should be insured though. They didn't fail because of mark-to-market losses - they failed because of those plus the fact that effectively all their deposits were uninsured. We already know how to fix this if we want to.
Without high interest rates they can’t get inflation under control. It’s down a bit but elevated from before. The next crisis will drive it back up.
However, lots of people/institutions can’t function at these rates. Quite frankly I’m stumped how anyone buys a house at these rates.
Thank-you for saying this. I am old enough to remember the S&L scandal, and I'm not even American!
Money creation gone wrong part....?
Is this a good place to ask stupid question? I am new here, I know nothing about these issues, hoping to find some help...
I understand the first step-- SVB buys government bond/mortgages because they require a smaller capital cushion. In the absence of "mark to market" the capital cushion is smaller than it should be. Right? But I don't understand the precise steps that take us from
a) rising interest rates
b) declining value of bonds
c) bank run
"Roughly speaking, if the bank paid $100 to buy a long-term bond with an interest rate of 2-1/2%, and now the interest rate on a comparable bond is 5%, the bank’s bond is worth about $50."
My first thought was "no way." After Googling I find that a quick approximation is the change in interest rate x years duration.
2.5%x20 years = 50%
Is this accurate for large changes?
No.
20 years it's 68%
30 years it's 61%