Skip to content

Archive for

Want to See It, Kinda See It, at Least I See the Numbers

I owned a few AI related names heading into this week – I had some PLTR, some INOD also some LPSN (which hasn’t worked, and I should probably dump).  I also own GOOGL of course.

The first two have been almost like placeholders, I don’t really trust them, but they keep my foot in the door just in case AI is as big as the hype.  I have kept my positions small enough so as not to do damage if it isn’t.  Because I haven’t been able to really “see it”, as they say.

I needed an ‘aha’ moment.  When I’m like oh, I get it, this is going to be huge.

As of today, I still don’t feel like I’ve quite had that aha moment.  But I’m closer to one after those NVDA results.

I just can’t ignore results like what NVDA released Thursday.  I mean to project nearly $4b more revenue in fiscal Q2 than analysts had expected is crazy. Something is going on here.  I still don’t know exactly what, what it means or who is going to benefit, but I have to pay attention here.

I remember the commodity boom of 2003-2007.  There was lots of poo pooing of the doublings that happened in a couple months at the start of that boom.  I’ve told the story before but I remember Aur Resources went from like $2 to $4 in a month or two after basically doing nothing for a decade.  And then I bought it, even though it made me ill.  But it turned out it was just the start of a 5 year move because of China.  Aur eventually got taken out over $40.

I don’t want to discount that possibility here.  That this is real and just the start.

I was actually short a little VICR heading into the NVDA results.  I know VICR well, I trade the stock back and forth because of that, and their outlook from the last earnings call was just so awful, while the stock had gone up to high-$40s, so it seemed like a decent swing short back to the low $40s.   But on Wednesday night I quickly covered and flipped long.

NVDA’s call not only said their H100 GPU was selling strong, but all their GPUs were selling strong and VICR has power conversion products on those other GPUs as well as other companies AI geared GPUs (VICR is also potentially going to be in the H100 at some point but that is a long story that would be an entire post).   I also bought some PATH, bought some SNOW and took smaller starters in IBM and HPE.

I don’t know if I will keep all those names.  I really don’t know what I’m doing here, at the moment there is an element of stabbing at the dark looking for winners, which is why all these positions are small.  I tend to act first and then look deeper because I don’t have the time.  I add and quickly subtract if I dig in and it doesn’t make sense.  I’ll probably have sold 2-3 of those names by mid next week and maybe added others as I figure out what I really want to own.

I still don’t really see “it”.  “It” being, how is this going to play out.   But what I do “see” now is how big the numbers can be.  The NVDA results were just so impressive.

I mean I look at the NVDA results.  The EPS forecast from analysts for their fiscal Q224 (current quarter) nearly doubled overnight from just over $1 per share to $2 per share.  It feels absolutely crazy to say this, but I don’t even know if NVDA is actually all that expensive here – and that is after the stock has gone up 4x in the last 6 months!  I mean if they keep beating estimates, what if they are doing $12 or $14 EPS in their fiscal 2025 (YE January 2025). Estimates for F2025 right now are at $10 but were $6 just 3 days ago!  $390 divided by 14 is 27x EPS.  Its not ridiculous.

So I see it, I see the numbers.  What I don’t see is how this impacts everyone else.  I can’t wrap my head around it yet.  Maybe the impact to IBM is negligible.  Maybe PATH is a loser in all this (LPSN seems to be).  Maybe PLTR is just full of BS with comments like this (from their Q1 earnings call):

in the last 20 years, there’s never been a development like this. You have a technology that will allow you to outproduce, change the margin of your company, understand your business, react on the battlefield quicker, predict things on the battlefield in a way, collapse your enterprise so that the top and the bottom actually work together, preempt attacks, create a software that is so obviously dominant that adversaries quiver and scurry away instead of attacking us or our allies.

I bought PLTR right after reading this earnings transcript.  I remember saying to my wife, I don’t know if these guys are full of shit or not, and they very well may be, but if anyone believes what they are saying, the stock is going up.

I was sitting at the kitchen table that night and i said to her: “I have no idea what is going on here or why this is AI is a big deal but there are companies losing their shit over this.” True story.

Flash forward to today and I still have no clue what to believe. Which is why its difficult to write this post.  You want to write to inform.  But I am writing to say, “I think there might be something going on here.  I don’t really understand it.  TBH it kinda triggers my BS meter.  But I’m wading in further”. 

Great thesis.

You gotta listen to the market.  Right now, the market is saying there is something going on, and this feels like it could be the start of one of those secular changes.

You can easily call these moves a bubble or irrational or a blow off top.  The chart of NVDA certainly looks like that.  But sometimes these things are legitimate secular moves.   Again, I can’t shake my copper analogy.  If it hadn’t been for Donald Coxe looking at those same copper stock charts up 50%-100% in a few months and saying you have to buy these things now, I never would have had the courage or insight to do it.

I just want to have that broader aha moment for what AI can do for so many stocks to go along with what I had (too late) this week for NVDA.   Unfortunately, its not as obvious as a “1 billion people moving from china countryside to cities” where the thesis just jumps off the page at you.  But that also means opportunity.   My suspicion is that some of the winners are going to be losers and vice versa.  That should mean lots of opportunities for the next while.  If I can only figure them out.

Well That’s Surprising

I’ve spent the last 3 posts looking at banks.  And it’s not going to stop.   Because this is what I find interesting right now.

From post #1 to post #3 my view has changed significantly.  I went from ‘OMG this is a disaster’ to ‘you know some banks should hold up okay’ to ‘its really weird how big the gap there is between what the stock price is doing and what bank management is guiding to’.

All the time I scratch my head and wonder how it will end?

My next installment is a ‘well that’s surprising’ post.  That is to say: ‘well that’s surprising’ that one of the banks at the center of the maelstrom actually doesn’t look like a complete disaster even when I model it out in very stressed scenarios.

The bank I’m talking about is Western Alliance Bancorporation.  Its right in the middle of this.  RIGHT in the middle.  It was hammered down to single digits when Silicon Bank went belly up, it was hammered down again when First Republic went belly up, it was bought by Chris Whalen at the peak of the crisis and then sold by Chris Whalen at what is now being called the Whalen bottom.  You couldn’t be more in the middle.

Now I just want to preface all this by saying I don’t own WAL.  I have no intention of owning WAL.  My conclusion is that they are at best walking a tight rope.  And they aren’t all that cheap compared to other banks out there given the potential for further distress.  So don’t turn me into a WAL bull.

But the way the stock performs (absent the rally this week) you’d expect things to be extremely dire.   What I found instead was that things were more like ‘not great’.

Like I said in the last post, as much as anything I’m putting my thoughts down for my own record – so I can go back and say ahh, that’s what I missed.

Onto WAL.

WAL is interesting for a few reasons.  First, this is not a security heavy bank.  They aren’t like Silicon Bank with massive MTM losses on Treasuries and mortgage-backed securities.

Second, WAL originates loans with a pretty high interest rate.  The average rate on WAL’s loans in Q1 2023 was 6.28%.  Compare this to say Keycorp – which had 5.04%.

Of course, that means that WAL loans should be riskier.   And they probably are.  But right now, with the economy still apparently trucking along, that riskiness isn’t bearing out into much higher loan losses.  So in a weird twist of fate, banks like WAL look better, at least on paper, because their higher rate loans can cover the rise in deposit costs better.

For WAL, the combination of high-rate loans and a relatively small portion of (low-rate) investment securities means that their net interest margin was not terrible in Q1, even though their deposit cost rose a lot.

Now that is going to get way worse in Q2 and beyond because WAL has seen an outflow of deposits since. To plug the hole they are tapping higher cost sources of deposits (or the Fed) to replace them.So far, no surprise.  This is where you would expect disaster.  Higher deposit costs, a collapse in net interest margin, a general mess.

What surprised me is that when I look at WAL and put some fairly significant stress on their deposits, I don’t get a complete disaster.   It is not exactly rosy, but it wasn’t nearly as bad as I thought it would be.

My assumptions started with management guidance.  Take that FWIW.

I tried to stay consistent with WAL’s loan growth and deposit growth estimate.  These are two things that management can control.

I did not follow management guidance for net interest margin, efficiency or pre-provision net revenue estimate because I wanted to assume much higher deposit costs, which throws off all of these metrics.

Essentially, I am assuming that WAL can build back their deposits but at a painful price.  The key assumption I’ve made, and really this is just a guess, is that WAL has a deposit beta of 80% by the end of the year.  That means that if the Fed keeps rates at 5.25%, WAL will have deposit costs of 5.25% x 0.8 = 4.2%. Which is pretty high.

Below is what I get for earnings.  Now this is kind of a pseudo-run rate on earnings.  I’m not trying to come up with a 2023 number or a 2024 number.   It is more of an annualized steady-state look of where things could be at YE in a distressed scenario.

It is not great, but it is not terrible.  Of course, to put it in context, WAL made almost $10 EPS last year.  So it is a massive drop in profitability.  But they are still profitable, which is more than I would have expected.

Next, I assumed the absolutely apocalyptic case where deposit costs go right up to the Fed Funds rate (Beta = 1).  In this case, as you might expect, EPS goes negative, but surprisingly it is still less than a $1/share loss.  I would have expected much worse?

Why are these estimates are better than I would have expected? I think it comes down to WAL’s high interest rate loans that can cover the deposit costs (and most of the other non-interest expense) even when those deposit costs are through the roof.  WAL also has a strong mortgage business that delivers fee income (their non-interest income was $100 million in Q1).

There are plenty of caveats to consider with these numbers. For one, I am not thinking about credit quality at all here. If WAL’s loans start to go south, things would fall apart quickly.

WAL has also entered this hyper-sensitivity zone where the numbers are extra sensitive to changes in assumptions as you get closer to break-even. It becomes a precarious tight rope.  For example, if non-interest bearing deposits drop to 20% (instead of the 25% I assumed), EPS drops to $1.90 and -$1.37 per share for the two scenarios.  It doesn’t take much for things to spiral down.

Which is why I am absolutely not saying WAL is a stock I am looking to buy.  I mean it would be trading at >10x earnings on my 80% beta stressed case. That situation is not implausible. There are plenty of banks that have lower PE’s without all the distress and certainly not the battleground element.

Nevertheless, it strikes me as interesting that even with a battleground stock and what should be some very extreme assumptions, I don’t get a complete disaster of a result. Surprising!

Are all the banks the same?

Thursday I tweeted the following:

It seems like pretty much everyone is saying that the regional banks are hooped.  You’ve got Chris Whalen saying that ALLY and COF are next.   You have Hugh Hendry coming out on Bloomberg saying that its going to get “real bad” and that he suspects “fed officials are considering a lock on bank deposits”.  Jim Bianco said on the Forward Guidance podcast that “everyone is demanding their money back right away”. 

This is almost no differentiation made when banks are talked about.  They are simply referred to as “the regional banks”.   All 4,000+ of them (community banks included).

It should thus be no surprise that this last move down has not discriminated much.  It has just been an across the board wipe out.

But are all banks facing the same stress? 

Here’s what I did last week.  I took a bunch of the larger regional banks that provided FY 2023 guidance with their Q1 results and I roughed out the numbers of what their earning might look like.

I kept reading on Twitter that the move in the regional banks was justified because regional banks earnings are becoming so impaired.

So surely this must mean that regional banks guidance for 2023 has come down dramatically?  I mean they reported only two weeks ago.  And this after the blow-ups of all but FRC.  Some of this impairments must be in the guides, right?

I was curious how bad it would be.

I didn’t use any crazy assumptions.  I pretty much stuck with A. what the company guided to and B. what first quarter results were.

Here is what those results looked like for two fairly large regional banks that have been hit quite hard in the last week and even harder in the last month and a half: FITB and KEY.

Fifth Third Bancorp

Starting with FITB, here is the guidance they gave on their Q1 call:

Loans up, net interest income up, non interest income up, expenses up. 

FITB gave the following guidance for deposit betas.   I used 45% for my average, which I ballpark is about inline with their worst-case scenario where beta reaches 49% by the end of Q423:

FITB also gave us their average interest assets and liabilities table on Page 21 of their Q1 Earnings Release.

From this information, in the spreadsheet below I essentially reconciled guidance and Q1 results to come up with what earnings would look like to be inline with it all.

I played around with loan and security yields to get something close to the net interest margin growth they guided to (note I ended up with 7.1% vs their 7-10% range).

Bright yellow cells are guidance or estimates.  The grey cells are used for comparison but not in the calculations.  The cell described at “Additional Interest Expense due to beta + mix” is just the FY 2023 deposit cost less the Q1 2023 deposit cost annualized.  I added this calculation to see just how much the additional deposit cost of rising beta and lower non-interest bearing deposits is to each bank.

It is a curious result.  All in, FITB earnings come out at $3.55 EPS.  Their NIM declines slightly even though their interest-bearing deposit costs rise from 1.76% to an average of 2.36% (because their loans and securities continue to earn more as well).

In addition to the guidance, I assumed provision for loan losses increases as the year goes on (averaging $184mm per quarter for the rest of the year).  I also assumed non-interest-bearing deposits decline from 31.6% to a 27% average – which would be consistent with a year-end number of about ~25% NIBs. 


I did the same thing for KEY.  Now KEY had a bad Q1.  They missed on their estimates and lowered their guide.  When I set out to work through what this meant for 2023, I thought the result may be quite ugly.  After all, consider Mike Mayo’s comment for KEY during the Q&A:

It’s rough when an analyst is calling you out like that.  This was KEY’s outlook:

They lowered net interest income and increased their tax rate.

KEY gave us an estimate of deposit betas “peaking in the low-40s”. 

KEY also gave one other data point that is interesting.  Below is the impact on net interest income of the rollover of short-term securities.  This impact is fairly low for 2023 ($79 million total) but it increases a lot going into 2024.  I think it is interesting because much of what we are told implies that the banks will only face rising costs and that they have nothing to offset those costs.

I included this in the line Forecasted Pickup from Treasuries and Swaps.

And here is what KEY’s forecast looks like.  Again, bright yellow are their estimates, grey are used for comparison not in calculations.

Surprising.  KEY earnings come in at $1.57 per share.

I don’t know, I mean I welcome anyone to go through these numbers and see where they might be wrong.  But they are essentially built from management guidance so unless I have a math error, there isn’t a lot else that could be going on.

So I don’t know, what am I missing?  Why did KEY fall 10% on Thursday into the $8’s, FITB fall 7% to nearly a new 52-week low?  All I saw on twitter was that the regional banks are in trouble and the fall in the regional banks is justified and regional banks are facing deposit runs and regional banks business model is fundamentally impaired.

There are really two things that can wrong with these two banks:

  1. Their deposit costs could come in much higher than they are estimating.
  2. Their loan book could have far more defaults.

On point #1, this has to be the biggest unknown.  Will deposit costs go up even more?  Will betas (measured as deposit costs as a percentage of the Fed Funds rate) far exceed historical norms? This has to be what a big part of what the market is pricing in.  After all, deposit costs rose a lot in Q1.  Maybe this is just the start.

On the other hand, Q1 may have been the peak for beta velocity. Rate increases have slowed. Beta lags, which means Q1 was pricing in much of the extremely fast rate velocity of late-2022. Will it really just keep accelerating?

I don’t know, but I wanted to look at what happened if beta got really bad. If I add 20 points to the beta of KEY and FITB, bringing it up to 62% and 65% respectively, earnings at KEY go down to $0.61 per share and FITB down to $2.25 per share.

That is a steep decline. 

But a 60% deposit beta would be extremely high compared to historical standards.  I went back and looked at what banks were saying when they were modeling beta early in the cycle.  KEY was saying that 30% was roughly the historical average.  FITB said the previous cycle (which did have a much lower terminal rate) was 38%.  We are already looking at numbers above that and if it gets to 60%, it would be way, way above that.

Could it really go that high? I don’t know. But this unknown has to be the big risk right now. If there is a second order comment to be made it is that if beta goes that high, the bank earnings get very sensitive to small changes. For example, KEY with 62% beta gives me 61c EPS. At 60% beta its 70c. At 65% beta its 46c.

The other thing that could go wrong is the loan book.  KEY and FITB aren’t particularly exposed to commercial real estate (8.9% and 8.4% of total loans respectively).  Some regionals, particularly the smaller one’s, have CRE exposure that is so large that really you have to take their earnings with a grain of salt if you think CRE is in trouble.

But if CRE becomes such a big problem that KEY and FITB are stressed by it, there are 3 out of 4 of the banks in the country, including some bigger one’s than these two, that will be in big trouble.

We will see.  Going into last week I had a few shorts on regional banks that were A. exposed heavily to CRE, B. had very high and rising deposit costs and C. saw their NIM shrink a lot in Q1.  On Thursday I covered most of those and I went long KEY and MTB (I am already long PNC) for a short-term trade.   

While this feels like anything but a fat-pitch (I’m not convinced that the worst if over given where the economy is and where it may cause inflation to go and I really have no clue how high deposit costs go over the long-run) it is a trade that worked out at least for Friday.   We’ll see if that continues.

The thesis is that A. management isn’t lying about their outlook based on where they see things now, B. that shorts have gotten complacent and are just shorting the regional bank index (KRE) which is dragging down all banks when not all banks are in the same boat. C. nothing in the economy or in their business is deteriorating as rapidly as the stock prices suggest it is.

What I Learned from Reading Bank Reports

Every quarter I read through a bunch of bank reports and earning calls.

Honestly, it’s a little dry.

I already wrote about what happened last fall when I went through the Q2 2022 reports.  In September I had a bit of a WTF moment as I saw:

  1. The tangible book value of almost every bank collapse – some by 30-40%
  2. No one seeming to care that this had just happened

I wrote at the time:

Again, I don’t know what to make of these OCI hits. Do they matter? I’m not sure. They certainly make the bank look more expensive on a book value basis. Post the OCI hit, the bank is closer to running into regulatory trouble if something goes wrong with its loan book. OCI hits don’t impact earnings, though as I said they might impact the ability to grow earnings in the future.

Just to be safe, I’ve stayed away from banks with big securities portfolios that are taking big OCI hits. Funnily, its either the very small banks or very large banks that aren’t having to do that. The small banks don’t own a lot of securities. the largest one’s (like C, BAC or JPM) seem to be better at hedging their exposure.

I was just kind of like – does this matter?  And I concluded that it probably didn’t matter.  Unless someone decided it mattered.  And then it might matter a lot.  So it was best to just stay away, just in case.

This quarter (q1 2023) I did the same thing I always do except even moreso.  I’ve read through a lot of reports and calls.

I went into this quarter thinking there might be another big “Holy Cow” moment from hearing what the bankers had to say.   But you know, there wasn’t.

Instead, the problem I think is more insidious.  Insidious – harmful but in a gradual, subtle way that takes time to show what is really going on.  It is the frog slowing boiling and not knowing it analogy.

The problem, which we all know, is that short term rates went from 1% to 5% in a year and banks can’t adjust to that fast enough.  I mean that’s really it.  Everything else is a consequence of that.

Consequence #1 is that deposit costs are just dripping upwards every day.  I pointed this out in my post on PNC.  Some banks saw 50 bps increases to their deposits in Q1.  Some saw as much as 100 bps increases.  But it’s going up for all of them.

The loan books can’t go up as quickly.  In part that is because they already have moved up, which pushed this problem from last year to this year.  In part it is because a bank can only roll over at most 20% of their loan book in a year.  And also only in part because some of the loans are variable rate which do go up and down with short term rates.

There are lots of factors.  Things can be very different depending on the bank.  What really matters is the net interest margin (NIM) – the spread between what the bank lends at and what they pay for deposits.  In Q1 some banks saw their NIM fall quite a bit as deposits rose and their loan and security book didn’t rise with it.  That isn’t a good thing.

But not all banks saw this.  For some NIM just moved down a little.  And some banks that had very “sticky” deposits or a lot of variable loans actually saw their NIM rise.

But most saw it fall at least a bit.  And it will fall again in Q2.   Now again, another nuance: some banks will try to offset this falling NIM with new loans.  And you heard that on various calls.  We’ll grow our way out of it.

But other banks can’t do this so they didn’t really talk about it.  These banks have a lot of loans on their balance sheet already (loan to deposit ratios of >100 while FRC, that bank that just failed, was actually only 95 in comparison) and so they can’t make a lot of new loans.  They are kind of stuck.

If you see a bank with a smallish NIM to begin with (<3% or yikes, <2.5%) and high loan/deposit ratio, that’s not a good thing.

Another thing I noticed is that most of these smaller banks own a lot of commercial real estate loans.  Bank after bank has 30% or 40% CRE.  The outliers have 90%+.  Very few banks don’t have significant exposure to CRE of one type or another.

Now not all of that is office.  There is industrial, retail, multifamily.  The one’s that have office are in probably the most hot water, and you can see that because every bank that didn’t have big office exposure told us they didn’t have big office exposure.

You get a hint of the problems with office exposure every once in a while.  For example, consider the following exchanges from one bank call:

And then later in the call:

As Seinfeld said, that’s a big matzah ball! 

So there’s office.  Its not great.  There are pockets of multifamily that aren’t great if you can’t raise rents (like in NYC or hmm, there was some city in Canada that did something like this, it will come to me…).  There are malls.  Lots of stuff that is a bit wobbly.

But I don’t think in and of itself these CRE loans would be a problem.   The banks would manage out of it.  It is like I said, every bank is different.  Some are more exposed, some less, a few will stumble, most won’t.

The bigger issue is Consequence #2 of the rise in rates.   And that is that the loan books of all the banks, every one of them, is underwater.

Any bank that wrote a fixed rate loan between 2020 and 2022 is underwater now.  Its not really their fault, its just the way it is.  They made the loan when fed funds were 0% and treasury rates were 1% and now rates are higher and so the loan is worth less.

Now NO BANKS talk about this on their conference calls.  There wasn’t one bank that talked about the mark to market of their loan book.  Because its not a good thing.  It’s a lot like OCI was last fall – better left unsaid and it isn’t a problem unless someone decides it is a problem.

Which brings up one interesting thing I did a couple weeks ago.  I went way all the way back to the summer of 2008 and read what a bunch of banks said on their Q2 2008 conference calls.

This was like 3 months before all hell broke loose.   And you know what?  They didn’t have a clue.  Or at least, they weren’t going to say it.

The best you got out of bank comments was a bit about tightening conditions at the edges, a little weakness here or there.  Seems like housing is a little weak.  Absolutely no blazing red flags.

So you can’t just listen to what they say.

I suspect that one of the reasons the banks have fallen out of bed this week is because when FRC went belly up over the weekend and was bought by JP Morgan, the loan book was bought for about a 13% discount and that is with a loss sharing agreement.

Think about that for a second.  The entire loan book of FRC is worth 13% less than the number on the balance sheet.

What would that mean if you had to sell the loans at fair value for every other bank tomorrow?

Again, its nuanced – some banks would be fine, others would be not so much.  But I’ll tell you, when this whole thing started back in March I went through the 10-Ks of about 30 banks and what I saw, where it was reported, was an average loan book down ~10% when it was marked to market (this wasn’t quite hidden at the end of the 10-K so you really had to hunt for it, but it was often disclosed).

It was pointed out somewhere that there were about $25-$30 trillion of loans made in 2020 and 2021 (I’m not sure if that is just banks or all financial entities, I’m getting this second hand but I think that number is ballpark right), when rates were super-duper low.  If you do the mark to market loss on that at a 15% haircut, that is in the range of $4 trillion of mark-to-market losses, just sitting there.

Now does that matter?  It doesn’t really matter if no one cares.  Its all held on the balance sheet of banks who don’t really want to sell it so if they don’t have to sell it then who cares.  But if it starts to matter, it would REALLY matter.

Which is why I just need to be a little bit careful here.