All Out May 26th: More debt ceiling, AI spooks markets, pension funds and structured products
It seems the debt ceiling will be taken down to the wire. Treasury secretary Janet Yellen has said the X date (when the government will run out of money) is roughly 1 week out. This is just an estimate and it may be a bit later. However, we don’t want to find out when it is. This is already playing into market distortions. The S&P 500 is trading just shy from its YTD high, volatility isn’t all that elevated (VIX is at 20), but there are certain T-bills which are being affected. The T-bill maturing on June 1st and 6th were trading around 6.2% last Wednesday, supposedly they hit as high as 7% in volatile trading. To put this in perspective, these yields are higher than some short dated junk bonds. As reference 1 month T-bills are trading around 5.7% and 3 months around 5.3%. So markets are worried something might happen and many participants don’t want to hold the bonds in question.
We’ve said this many times, but the debt ceiling is dumb. It doesn’t accomplish anything other than market stress. Eventually it will be raised and it won’t actually do anything to contain spending or lower the budget deficit (we don’t think it matters all that much just yet, but that’s a topic for another day). We heard a podcast today with another potential solution to the debt ceiling we hadn’t heard before. We talked before about the trillion dollar platinum coin and high interest rate debt. There are 2 other solutions. The first one is invoking the 14th amendment which says, amongst other things, that the validity of the public debt” of the U.S. “authorized by law … shall not be questioned”. So there’s an argument that the government could use this to keep on spending as to not violate the constitution (this sets a dangerous precedent though).
The other solution highlights the stupidity of the problem. Supposedly (we haven’t read it) the text that limits the debt specifically says that the face value of the debt issued cannot be more than X (this is why the high interest debt argument works). The other solution is to issue perpetual bonds, so called consol bonds. These bonds are meant to never be redeemed therefore, they do not have a face value. In theory the government could issue as many perpetual bonds as it wants and keep on spending (again another dangerous precedent). These bonds have been issued before, but Treasury probably cannot get their act together in time to issue the bonds before they run out of money. Again, the debt ceiling is a dumb gimmick that doesn’t provide any value. Hopefully we can stop writing about this soon (until the next time it’s reached).
How Fake AI Photo of a Pentagon Blast Went Viral and Briefly Spooked Stocks
Pretty scary thing happened in markets a couple of days ago. With the advent of chatGPT, Dall-E, etc, it’s fairly easy to come up with fake images, voices, and even videos. Apparently a “verified” Twitter account tweeted an AI generated image of an explosion at the Pentagon. A soon as this happened markets quickly dropped and the S&P 500 wiped out $500 billion of market cap in a couple of minutes. When the image was found out to be fake, markets quickly recovered.
This reminds us of when Twitter changed how the blue checkmark worked from being verified to just having to pay for it. Someone created a fake Eli Lilly account, paid for the verified checkmark and tweeted out that insulin was now free. The stock quickly dropped 4%. This was clearly a mistake and Twitter fixed their blue checkmark process to not have this happen again (it did happen a couple of more times though).
The scary thing with AI is that pretty much anybody with a computer and some basic AI skills can start generating fake images, voices, videos. How will people know what’s fake and what isn’t? What will the effect be on markets? What happened with the pentagon picture was probably driven by automated trading. Can these algorithms be fooled with other AI images, text, voices? Supposedly the flash crash of 2010 was caused by a single trader in the UK trading out of his basement. What happens if there is a coordinated hoax (attack?) of fake images, videos, news reports etc. Will these algos drive markets down and create another flash crash? Will trades be broken up like they were in the flash crash? Clearly we don’t know the answers to these questions but something to keep in mind.
New York pension funds take different approaches to setting rates of return
Pension funds are vehicles designed to save for retirement. Workers put money aside, employers pay in, and the pension fund invests the funds in order to pay out benefits when the employees retire. A key metric in a pension fund is its funded status. This basically measures how much money they have in relation to their liabilities (how much they need to pay out in the future). The liabilities are fairly easy to know, you have X amount of workers, they retire at certain age, benefits are Y and you can calculate how much you need to pay out and when. The asset side of the equation is pretty simple as well, you basically see how much money you have in the pension funds accounts. However, these values are at two different points in time. Assets are today, liabilities are in the future, these two are not directly comparable because of the time value of money (a dollar today is worth more than a dollar in the future).
In order to make these two amounts comparable, pension funds assume a rate of return on their investments. Therefore they can say, we need to pay out $100 in 10 years, today we have $35 and we assume they will grow at 10%, so in 10 years those $35 will be worth $90.7. Therefore, in this hypothetical example the funded ratio would be 90.7%. In 10 years they will still be missing $9.3 dollars to meet their obligations. There are three ways of fixing this problem. You either deposit more money today, you pay out less in the future, or you earn higher returns on your investments. Options 1 and 2 are not popular with retirees since they will have either less money today or less money in the future. So the incentive is to generate higher returns. However, one thing is actually generating higher returns, the other is simply assuming you will generate higher returns. There is no set way of estimating expected returns and these expectations will vary greatly with market conditions. For example, today we can assume much higher expected returns for bonds than we could 2 years ago (rates today are ~5%, and 2 years ago they were 0%).
This is a reason why pension funds have flocked to private equity. Private equity advertises (falsely we would argue) higher returns than public equities (again these numbers aren’t comparable but that’s a topic for another day). Therefore pension funds can plug in this higher expected return into their models and magically improve their funding ratio. There’s also another thing coming into play here which is career risk and the principal agent problem. The pension fund’s CIO isn’t the owner of the capital. Therefore, their interests might not be aligned with the retirees. The objective of the CIO is to keep their job. In theory they do this by generating strong returns for the pension plan. However, the pension fund is a long term investment vehicle, which has a time horizon measured in decades. Most employees probably won’t stick around that long. Therefore, if they engage in unconventional thinking, which might be most beneficial in the long term, and if it goes wrong in the short term they won’t be around to benefit from the results. The problem here is that the decision will be reversed when the CIO is fired, which is possibly the worst possible time to get out of the investment. We saw this with CalPERS (the largest pension fund in the US), with their tail risk hedging program. It had been implemented a couple of years back, lost money most years (as it is expected to do), and was removed right before the Covid crash; a move which cost them billions.
Structured Products: Right Idea, Wrong Execution
We’ve written about structured products before. They are very popular with investors, but are often a subpar structure for investors. A structured product is essentially a debt instrument issued by a bank in which they promise certain return characteristics. For example a structured note can promise to pay a coupon of 15% if the S&P 500 doesn’t drop below 20%. If it does drop below 20% then the investor loses the same amount as the S&P 500. This is just an example but it’s to illustrate what what note can look like. Structured notes essentially reshape the payout profiles of different assets. Banks that issue these don’t take the other side. They hedge out their risk and they gain from the embedded commissions in the product (which are large).
Structured notes have several problems:
- Credit risk: banks issue these as unsecured debt instruments, so if the bank fails, the note goes to zero. The problem with this risk is that the investor is not being compensated for that risk.
- Lack of liquidity: it is hard to find information for many structured notes since they are not publicly traded.
- Expensive: banks charge hefty fees for these products and these fees are often hard to find and understand.
- Risky: many structured notes are pitched as fixed income replacements. Their payoff is oftentimes called a “coupon”, like a bond. However, many of these “coupons” are contingent on the underlying asset not dropping too much. And if the asset does drop more than X amount the the investor may lose 1 for 1 with the asset. So for example notes can have 15% upside at the risk of total loss.
- Complex: notes oftentimes can have several underlying assets, knock in or knock out features, etc. They are hard to understand, therefore hard to price and hard to model their role in a portfolio.
So what to do if you like structured notes? Well typically these payoffs can be replicated by simply buying treasury bills and buying and selling put and call options. The replication approach has several advantages, such as more liquidity, more customization, cheaper, and you can manage them before expiration. If you want to hear more about how this can be done, contact us!
