What We’re Reading (Week Ending 19 March 2023) - 19 Mar 2023
Reading helps us learn about the world and it is a really important aspect of investing. The legendary Charlie Munger even goes so far as to say that “I don’t think you can get to be a really good investor over a broad range without doing a massive amount of reading.” We (the co-founders of Compounder Fund) read widely across a range of topics, including investing, business, technology, and the world in general. We want to regularly share the best articles we’ve come across recently. Here they are (for the week ending 19 March 2023):
1. Poignant Twitter thread on the importance of having purpose in life – Mark McGrath
This is a story about my dad. My dad grew up the youngest of four siblings in Quebec. He, his siblings, and my grandparents moved to Vancouver in the 70s, and my uncle opened a tile store. My dad worked for him for a while, then eventually opened his own store.
He was a relentless entrepreneur and a good father. He was shrewd and pennywise. I used to joke that he would split 2-ply toilet paper to save money. But he was also a savvy investor, and he did well in his business.
He didn’t care about tile and saw the business as a means to an end – a way to build wealth and retire. He was laser-focused on this goal.
He was fit, active, and a traveller. He was a scratch golfer and swam 80 laps at the pool three times a week. He was also a black belt in karate and extremely disciplined…
…One day he told us he had sold his business and was retiring. We were thrilled. All he wanted to do was retire so he could keep travelling, golfing, swimming, and enjoying his life. He booked a two-month trip to Asia to celebrate. He was 58. And then it all went downhill.
Within a month of returning from his trip, he was back working for the guys he sold his store to. He didn’t need the money – he just missed his store and his friends. His best friend was his first employee – a man he had hired 30 years earlier. This worked out for a while.
But slowly, he started to change. After a few months of golfing near-daily, he got bored. And then he got depressed. He changed…
… Then he told me, “your father is dead.” I collapsed and remember only that I kept saying, “I had so much more to tell him”.
He had rented a car for some reason, drove it to the middle of the Lion’s Gate Bridge in Vancouver, turned on the hazard lights, and got out. Then he jumped. Two cyclists – one on the bridge, and one down below on the seawall – called it in…
…What I think happened is that my dad’s business became his identity. He was the tile guy. He was the guy that sponsored all of our sports teams. In a booming town, he was the guy you went to when you needed tile. He was the tile guy.
And when he sold his business, he stripped himself of his identity. Now he was a nobody. He lost his purpose, the very thing that made him who he was.
2. GPT-4 – OpenAI
We’ve created GPT-4, the latest milestone in OpenAI’s effort in scaling up deep learning. GPT-4 is a large multimodal model (accepting image and text inputs, emitting text outputs) that, while less capable than humans in many real-world scenarios, exhibits human-level performance on various professional and academic benchmarks. For example, it passes a simulated bar exam with a score around the top 10% of test takers; in contrast, GPT-3.5’s score was around the bottom 10%. We’ve spent 6 months iteratively aligning GPT-4 using lessons from our adversarial testing program as well as ChatGPT, resulting in our best-ever results (though far from perfect) on factuality, steerability, and refusing to go outside of guardrails.
Over the past two years, we rebuilt our entire deep learning stack and, together with Azure, co-designed a supercomputer from the ground up for our workload. A year ago, we trained GPT-3.5 as a first “test run” of the system. We found and fixed some bugs and improved our theoretical foundations. As a result, our GPT-4 training run was (for us at least!) unprecedentedly stable, becoming our first large model whose training performance we were able to accurately predict ahead of time. As we continue to focus on reliable scaling, we aim to hone our methodology to help us predict and prepare for future capabilities increasingly far in advance—something we view as critical for safety…
…In a casual conversation, the distinction between GPT-3.5 and GPT-4 can be subtle. The difference comes out when the complexity of the task reaches a sufficient threshold—GPT-4 is more reliable, creative, and able to handle much more nuanced instructions than GPT-3.5.
To understand the difference between the two models, we tested on a variety of benchmarks, including simulating exams that were originally designed for humans. We proceeded by using the most recent publicly-available tests (in the case of the Olympiads and AP free response questions) or by purchasing 2022–2023 editions of practice exams. We did no specific training for these exams…
…We also evaluated GPT-4 on traditional benchmarks designed for machine learning models. GPT-4 considerably outperforms existing large language models, alongside most state-of-the-art (SOTA) models which may include benchmark-specific crafting or additional training protocols:..
…GPT-4 can accept a prompt of text and images, which—parallel to the text-only setting—lets the user specify any vision or language task. Specifically, it generates text outputs (natural language, code, etc.) given inputs consisting of interspersed text and images. Over a range of domains—including documents with text and photographs, diagrams, or screenshots—GPT-4 exhibits similar capabilities as it does on text-only inputs. Furthermore, it can be augmented with test-time techniques that were developed for text-only language models, including few-shot and chain-of-thought prompting. Image inputs are still a research preview and not publicly available…
…Despite its capabilities, GPT-4 has similar limitations as earlier GPT models. Most importantly, it still is not fully reliable (it “hallucinates” facts and makes reasoning errors). Great care should be taken when using language model outputs, particularly in high-stakes contexts, with the exact protocol (such as human review, grounding with additional context, or avoiding high-stakes uses altogether) matching the needs of a specific use-case.
While still a real issue, GPT-4 significantly reduces hallucinations relative to previous models (which have themselves been improving with each iteration). GPT-4 scores 40% higher than our latest GPT-3.5 on our internal adversarial factuality evaluations:…
…GPT-4 can also be confidently wrong in its predictions, not taking care to double-check work when it’s likely to make a mistake. Interestingly, the base pre-trained model is highly calibrated (its predicted confidence in an answer generally matches the probability of being correct). However, through our current post-training process, the calibration is reduced…
…GPT-4 poses similar risks as previous models, such as generating harmful advice, buggy code, or inaccurate information. However, the additional capabilities of GPT-4 lead to new risk surfaces. To understand the extent of these risks, we engaged over 50 experts from domains such as AI alignment risks, cybersecurity, biorisk, trust and safety, and international security to adversarially test the model. Their findings specifically enabled us to test model behavior in high-risk areas which require expertise to evaluate. Feedback and data from these experts fed into our mitigations and improvements for the model; for example, we’ve collected additional data to improve GPT-4’s ability to refuse requests on how to synthesize dangerous chemicals.
GPT-4 incorporates an additional safety reward signal during RLHF training to reduce harmful outputs (as defined by our usage guidelines) by training the model to refuse requests for such content. The reward is provided by a GPT-4 zero-shot classifier judging safety boundaries and completion style on safety-related prompts. To prevent the model from refusing valid requests, we collect a diverse dataset from various sources (e.g., labeled production data, human red-teaming, model-generated prompts) and apply the safety reward signal (with a positive or negative value) on both allowed and disallowed categories.
Our mitigations have significantly improved many of GPT-4’s safety properties compared to GPT-3.5. We’ve decreased the model’s tendency to respond to requests for disallowed content by 82% compared to GPT-3.5, and GPT-4 responds to sensitive requests (e.g., medical advice and self-harm) in accordance with our policies 29% more often…
…Overall, our model-level interventions increase the difficulty of eliciting bad behavior but doing so is still possible. Additionally, there still exist “jailbreaks” to generate content which violate our usage guidelines. As the “risk per token” of AI systems increases, it will become critical to achieve extremely high degrees of reliability in these interventions; for now it’s important to complement these limitations with deployment-time safety techniques like monitoring for abuse…
…Like previous GPT models, the GPT-4 base model was trained to predict the next word in a document, and was trained using publicly available data (such as internet data) as well as data we’ve licensed. The data is a web-scale corpus of data including correct and incorrect solutions to math problems, weak and strong reasoning, self-contradictory and consistent statements, and representing a great variety of ideologies and ideas.
So when prompted with a question, the base model can respond in a wide variety of ways that might be far from a user’s intent. To align it with the user’s intent within guardrails, we fine-tune the model’s behavior using reinforcement learning with human feedback (RLHF).
Note that the model’s capabilities seem to come primarily from the pre-training process—RLHF does not improve exam performance (without active effort, it actually degrades it). But steering of the model comes from the post-training process—the base model requires prompt engineering to even know that it should answer the questions.
3. Bank Runs, Now & Then – Ben Carlson
Silicon Valley Bank, the 16th biggest bank in the country, was closed on Friday. It was the second-biggest bank failure in U.S. history…
…There is a lot to this story but it really boils down to an old-fashioned bank run. A flood of withdrawals from depositors destroyed the bank.
If everyone with a Planet Fitness membership showed up at the gym at the exact same time there would be chaos at the squat racks. It would be impossible for anyone to work out and the gym model wouldn’t work.
The same thing applies to banks. If everyone goes to get their money out on the same day, it’s going to be hard for a bank to survive…
…The SVB ordeal caused me to revisit my old copy of The Panic of 1907 by Robert Bruner and Sean Carr.
It’s a wonderful account of one of the biggest and most influential financial crises in history.
The Panic of 1907 would probably be more famous if it wasn’t overshadowed by the Great Depression just a couple of decades later.
It lasted 15 months and saw GDP decline an estimated 30% (even more than the Great Depression).
Commodity prices crashed. Bankruptcies exploded. The stock market fell 50%. Industrial production dropped by more than at any time in history up to that point. The unemployment rate went from 2.8% to 8%.
Trust in the financial system went out the window as banks failed left and right. In October and November of 1907 alone, 25 banks and 17 trust companies went under…
…Bruner and Carr laid out 7 reasons the Panic of 1907 was a perfect storm for bank runs and a massive financial crisis:
1. Complexity. Complexity makes it difficult to know what is going on and establishes linkages that enable contagion of the crisis to spread.
2. Buoyant growth. Economic expansion creates rising demands for capital and liquidity and the excessive mistakes that eventually must be corrected.
3. Inadequate safety buffers. In the late stages of an economic expansion, borrowers and creditors overreach in their use of debt, lowering the margin of safety in the financial system.
4. Adverse leadership. Prominent people in the public and private spheres wittingly and unwittingly may implement policies that raise uncertainty, thereby impairing confidence and elevating risk.
5. Real economic shock. An unexpected event (or events) hit the economy and financial system, causing sudden reversal in the outlook of investors and depositors.
6. Undue fear, greed, and other aberrations. Beyond a change in the rational economic outlook is a shift from optimism to pessimism that creates a self-reinforcing downward spiral. The more bad news, the more behavior that generates bad news.
7. Failure of collective action. The most well-intended responses by people on the scene prove inadequate to the challenge of the worst crises.
Again, not exactly like 1907 but this run on the bank seems to check all of the boxes in its own way…
…Two economists took a stab at explaining why bank runs happen and concluded they’re kind of random. Depositors are worried a financial shock will cause a lengthy liquidation so they pull their money en masse.
But what sets them off?
People being people, I guess?
4. Psychological Paths of Least Resistance – Morgan Housel
When faced with a problem, rarely do people ask, “What is the best, perfect, answer to this question?”
The more efficient question is often, “What answer to this question can I obtain with the least amount of effort, sacrifice, and mental discomfort?”
The psychological path of least resistance.
Most of the time that’s fine. You use a little intuition and common sense and find a practical answer that doesn’t rack your brain or bog you down with details.
Other times the easy answers lead you down a nasty path of misunderstanding, ignorance, and blindness toward risk.
A few paths of least resistances that everyone is susceptible to at some point:
1. The quick elimination of doubt and uncertainty.
Most people could not get out of bed in the morning if they were honest about how much of their future is unknown, hangs by a thread, or can be pushed in another direction by the slightest breeze. The solution is to eliminate doubt and uncertainty the moment they enter your head.
Uncertainty feels awful. So it’s comforting to have strong opinions even if you have no idea what you’re talking about, because shrugging your shoulders feels reckless when the stakes are high.
Life is complex, complex things are always uncertain, uncertainty feels dangerous, and having an answer makes danger feel reduced. It’s an easy path of least resistance.
If you were an adult in 2000 you probably had at least some vision of what the future would look like. Maybe even a vague vision of the next 20 years. But everyone was blind to 9/11, the 2008 financial crisis, and Covid-19 – the three risks that were both massive and unpredictable.
Then when those events happened people quickly moved to eliminate the uncertainty they brought.
Terrorist attack just happened? It’s definitely going to happen again, soon.
Recession coming? It won’t affect my industry and will be over by Q4 and interest rates will bottom at 3.42%.
Pandemic arrived? Two weeks to slow the spread.
No matter how wrong these answers might be, they feel better than saying, “I have no idea what’s going to happen next.”…
…5. The desire to supplant statistics with stories.
“People would rather believe than know,” said biologist E.O. Wilson.
I think another way to phrase it is that people desire stories more than statistics. They need a story they can tell themselves, not just a fact they can memorize.
Part of that is good. The gap between what works in a spreadsheet and what’s practical in real life can be a mile wide. This usually isn’t because we don’t know the statistics. It’s because real-life stories are so effective at showing us what certain parts of a statistic mean.
Part of it can be dangerous, when broad statistics are ignored over powerful anecdotes.
Government agencies published literally thousands of different economic data points, everything from unemployment to GDP growth to the historical cost of chicken legs, bone-in. It’s all free and easy to read.
How often do those data sites change average, ordinary people’s opinions about the economy?
It rounds to never.
What changes people’s minds and reaffirms their beliefs are conversions they’ve had with people close to them, social media, and cable news. Each is very good at telling stories especially when they provoke emotion or are easy to contextualize to their own lives.
When confronted with a pile of dull facts and a pile of compelling anecdotes, the anecdotes are always the path of least resistance for your brain to cling to.
5. TIP533: How The Fed Went Broke w/ Lyn Alden – Stig Brodersen and Lyn Alden
[00:01:28] Stig Brodersen: Well, thank you for saying so Lyn, and let’s just jump right into it. Today, I would like to talk about how the Fed went broke, but before we do, and perhaps to sort of like create a foundation for everyone, perhaps we can zoom out and if I can ask you to explain what is on the assets and liability side on the Fed balance sheet, and then perhaps we can talk about how that is similar to how a commercial bank running their balance sheet.
[00:01:53] Lyn Alden: Sure. So basically in a lot of regard, the Fed is very similar to a commercial bank. I mean, there are very important exceptions where it’s not, but in terms of the over, like arching details, it’s actually pretty similar. So if you look at a commercial bank for a second, they have assets and liabilities.
[00:02:08] Lyn Alden: The assets exceed the liabilities. That’s an important part of their solvency and their assets generally pay higher interest rates than their liabilities. Kind of the purpose of a bank is to, you know, borrow money at cheap rates and lend money with a little bit more risk and a little bit more duration at higher rates, as well as collecting fees and things like that along the way.
[00:02:27] Lyn Alden: And so for a typical bank, their liabilities are mainly their deposits. So basically when you deposit money in a bank, that’s your asset. It’s their liability and interest rates. They’re generally pretty. On the bank asset side, depending on the type of bank it is, they do mortgages, they do business loans, and they do credit card lending.
[00:02:47] Lyn Alden: They do all sorts of different types of lending, and those are ones that are generally a little bit riskier, higher duration, but they pay higher interest rates and so they can absorb some, you know, small percentage of defaults, build positive capital, pay [00:03:00] dividends, you know, fund their operations and maintain positive equity and positive capital.
[00:03:05] Lyn Alden: When you look at a central bank, it’s very similar, but there’s a couple different categories for their assets and liabilities. So their liabilities are, One bank notes, right? So physical cash and circulation is a liability of that country Central bank, and those are obviously 0% yielding assets, right? If you hold a dollar bill or a physical euro, you’re not getting paid interest on this.
[00:03:26] Lyn Alden: So that’s an obvious already good start for them, right? They have 0% liabilities there, but they have other liabilities that, for example, consists of bank reserves. So much like how we deposit money at a bank, and that’s our asset and their liability. Banks have to deposit their cash, their spare cash at the central bank, and that’s an asset for the bank, and it’s a liability for the central bank.
[00:03:47] Lyn Alden: And just like how a bank pays interest, a central bank also in, in many environments does pay interest on those reserves. And the reason they do that is because it’s an important part of how they manage their short-term interest rates. It basically presents a floor, right? If you can put reserves in the central bank, You know, and get, say 5% interest on it, there’s no reason why you would lend to anyone else at below 5% because you’re just taking on more risk and for less return.
[00:04:15] Lyn Alden: Right. And so that’s one of their important policy tools. And then there are other liabilities they can do, like reverse repos and things like that. They get more complex and some of those do pay interest. So that’s the central bank’s liability side. On the asset side, it actually looks [00:04:30] pretty similar to a commercial bank.
[00:04:31] Lyn Alden: They have things like treasuries, you know, the government debt of whatever country they operate in. So those pay interest. They also often have mortgage backed securities, right? So they have mortgage exposure. Obviously, these deals would differ around the world, but for example, a Federal Reserve has a lot of mortgage backed securities.
[00:04:46] Lyn Alden: These also pay interest. And then in some countries they’ll have things like corporate debt, or they’ll have things like equities. Those are generally considered less traditional types of assets for central banks to hold. But you see some like Japan kind of going that route. And sometimes, like the Fed and others will do that temporarily during crisis.
[00:05:02] Lyn Alden: Things like corporate debt. And in most contexts, the federal reserve’s assets are bigger than their liabilities and they pay higher interest rate than their liabilities. And it will then differ from jurisdictions. But usually the central banks operate like utility where it has to pay its excess profits back to the government.
[00:05:21] Lyn Alden: It doesn’t just keep building capital like a commercial bank would. Although in some jurisdictions they actually, you can publicly hold, you know, shares of a central bank and they will, you know, they could pay dividends, they could do things like that. Look at the Federal Reserve, so it’s not publicly held, but it is held by banks.
[00:05:38] Lyn Alden: They basically pay a small dividend to their owners. They pay their operating expenses, and then they have to send the rest of their profits back to the treasury. Right? And so it’s actually a source of income for the treasury, and it kind of makes it so that any sort of treasury is held by the Fed are effectively interest free because they are paying interest on them.
[00:05:55] Lyn Alden: But all these, a lot of these profits are getting sent right back to the treasury. The challenge in [00:06:00] recent months, really ever since September, is that the Federal Reserve increased interest rates so quickly and so significantly, and for the first time they got above the prior cycles high in terms of interest rates or at least, you know, the first time in, in decades we’ve had this kind of declining trend of lower highs in terms of interest rates, but they actually got way above that.
[00:06:18] Lyn Alden: And so they’re actually, their liabilities pay higher interest rates than their assets. And so obviously their bank notes are still paying zero, but their other areas, their bank reserves and their reverse repos in the fed’s case are paying a higher interest rate than their treasuries and their mortgage backed securities that in many cases are a longer duration.
[00:06:36] Lyn Alden: They’re fixed rate, they’re not adjusting upwards. They hold them from years ago, and so they have a mismatch. And so one is there, they’re no longer profitable. They’re not sending any more remittances to the treasury, and two, if they were a normal commercial bank, they would be on the verge of bankruptcy.
[00:06:51] Lyn Alden: So they’re months away from having negative, tangible equity, which is a normal bank would be bankrupt, but because of the central bank, they get to that. That’s where they have a very big difference. They basically get to just put a placeholder there that kind of is like an I O U. And so in the future, if they’re ever profitable again, then before sending more money to the treasury, They get to pay themselves back.
[00:07:13] Lyn Alden: So basically they’re losing money. They’re going in towards negative, tangible equity, but they’re filling that negative equity gap with IUs on their future income, which of course, for any private entity would be red flags over the place. Absolute catastrophe. You wouldn’t touch it with a 10 foot pole, [00:07:30] but it’s different if you’re the central bank.
6. Whose Fault is it Anyway – Michael Batnick
It has been 872 days since a bank failed in the United States. This was the longest streak on record. We’re now at day zero. Silicon Valley bank went down on Friday. Signature Bank last night. These are the second and third largest bank failures in history behind Washington Mutual during the GFC.
People are scared, mad, and looking for someone to blame. How did this happen, and whose fault is it anyway?…
…Blame the Fed
Three years ago, the fed appropriately took interest rates to zero as an economic meteor slammed into the Pacific Ocean. But two years later with the economy reopened and inflation running north of 7%, rates were still at zero. This made no sense then, and it makes less sense looking back on it. The fed was late to respond, and they compounded the problem by going from too easy for too long to too tight too fast. We haven’t seen a tightening cycle like this in the last fifty years.
A major thing that we didn’t anticipate as a result of these historic interest rates, at least I didn’t, were the ripple effects it would have at banks. According to Marc Rubinstein:
Between the end of 2019 and the first quarter of 2022, deposits at US banks rose by $5.40 trillion. With loan demand weak, only around 15% of that volume was channelled towards loans; the rest was invested in securities portfolios or kept as cash.
Banks invest their deposits in short-term bonds, for the most part. But even short-term bonds can have large unrealized losses when interest rates spike until the bonds mature. And bonds that have more interest rate risk are even more susceptible to large losses. All told, banks are now sitting on roughly $600 billion of losses in what are supposed to be among the safest instruments in the world. All because the fed went too far to fast.
Prior to aggressively raising rates, the fed kept interest rates at zero for too long which spurred excessive risk-taking. Venture capital was at the epicenter of this. Everything got funded in 2021 at a speed and size the likes of which the industry had never experienced. Who’s to blame here? Is it the fed for stoking the flames of speculation, is it the LPs for flinging money at venture funds, or is it the venture capitalists for saying yes to everything? The answer is yes.
7. What’s Going On With Treasuries? Silicon Valley Bank And The Incoherence Of The Federal Reserve’s (Lack Of) An Interest Rate Policy This Week – Nathan Tankus
The essential issue seems to be not so much “financial contagion” from the failure of Silicon Valley Bank (that’s how systemic risk is ordinarily understood.) Rather, it’s the implications of the Federal Reserve’s actions over the weekend. It is strange to see the Federal Reserve launch a facility commonly interpreted as a “crisis” facility using its 13(3) powers in the current economic situation(check out my Silicon Valley Bank piece as a refresher here). That’s because unemployment is low, inflation has been high and the Federal Reserve is raising interest rates. One interpretation of this event, consequently, is that the Federal Reserve is going to be lowering interest rates.
Yet, inflation remains above the Federal Reserve’s target. It would be quite an extraordinary situation in these circumstances to see the Federal Reserve lower interest rates at a time of elevated inflation. You can see the dilemma. Government securities dealers — those people who buy and sell treasuries every day — are as confused and unsure as you are about which way interest rates will go, in these circumstances. When bond traders are confused and unsure about which way interest rates will be going to this degree, treasury market issues result…
…The other layer to this are those “treasury liquidity strains” this all started with. When non-experts hear about “liquidity strains” in the treasury market, they tend to assume that means the U.S. treasury has to offer higher interest rates to sell securities. However, that often isn’t the case. In fact, these periods tend to coincide with falling treasury interest rates. In March 2020, liquidity in treasury markets worsened. Some short maturity treasury securities even experienced negative money interest rates. That means the situation was so uncertain, many actors were willing to pay such a premium that they would get less money back when the security matured then they paid for it. After all, losing a bit of your money is better than losing it all. To be clear, I’m not talking about “inflation adjusted” amounts. I mean literally, they paid 100 dollars, and got back 98 dollars.
So if “liquidity strains” don’t necessarily mean rising interest rates, what do they mean? They mean the price at which you can buy a treasury is further away from the price at which you can sell a treasury…
…Normally, bond traders have a pretty good sense of where interest rates are going to go. They are not always right (by any stretch). But this usually does not impact the differences between buying and selling prices that much. One way to think about this is, when bond traders are wrong they tend to be wrong slow enough and gradual enough that there is time for them to “catch up”. And generally, for the past 30 years if not more, when they are really wrong it’s obvious. So both buying and selling prices really jump. When Lehman Brothers collapsed, everyone understood short term interest rates were going to go to zero and stay there for a while. If inflation had been 6% when Lehman collapsed, the treasury market may have faced the same problems it’s facing now.
Faced with this uncertainty, the treasury market is getting less liquid. That means selling prices and buying prices are diverging even though interest rates are overall declining. But remember those bets that so many government bond market participants made? This is where they get hammered twice. Not only are interest rates going down when they expected interest rates are going up. In addition, “bidding interest rates” and “selling interest rates” are diverging, when these same bets often assume that the treasury market is liquid. In other words, embedded in these bets about the direction of interest rates are bets about the differences between bidding interest rates, and selling interest rates. In short, they also bet the spread would be small when the spread has been getting larger.
The final component of this on my radar is something that financial economist Daniela Gabor said over on Twitter:
“Why would you sell securities you can monetise at the Fed for par value?” This returns us to a more direct impact of the Silicon Valley Bank failure and the Federal Reserve response than the possible implications of that response for interest rate policy. The Bank Term Funding Program (again read Tuesday’s piece for the details) provides terms that are so overwhelmingly generous. That calls into question why any chartered bank (“depository institution”) who is allowed to access the BTFP would be selling treasury securities right now. What’s the point? You get a better deal handing it over to the Fed as security for a loan.
This makes sense for them individually, but it means suddenly trillions of dollars of treasury securities are not available for sale. Many fewer treasury securities “in circulation” must be having an impact on this liquidity situation. These banks would need a selling price that is much higher than the buying price, in order to be willing to sell their securities. This is a fluid and volatile situation, where the news coming out is confusing and fast paced. As a result, it will be months, maybe even years, before we have a good idea of how big the direct impact of this quasi-emergency facility was. Some other elements of this piece may be subject to revision once we learn more. But this is how I think the “state of play” looks today…
…It is no secret I have not been a fan of the Federal Reserve’s interest rate increases. However, if you are going to continue them then, when you announce the use of what is normally seen as a crisis facility, that should come with clear explanations of the implications for interest rates. Indeed, it’s maddening to even have to say this! Forward guidance is supposed to be what the modern Federal Reserve is all about! If they had said “interest rates will keep on increasing as usual, though we probably will be doing smaller interest rate hikes for the next three months”, then the implications of this situation would be clear. Treasury security interest rates would have rapidly adjusted without impairing treasury liquidity.
Disclaimer: None of the information or analysis presented is intended to form the basis for any offer or recommendation. We currently have a vested interest in Microsoft (parent of Azure). Holdings are subject to change at any time.
