What We’re Reading (Week Ending 01 September 2024) - 01 Sep 2024
Reading helps us learn about the world and it is a really important aspect of investing. The late Charlie Munger even went 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 01 September 2024):
1. Aidan Gomez, Co-founder & CEO @Cohere: What No One Understands About Foundation Models (Transcript here) – Harry Stebbings and Aidan Gomez
Aidan Gomez: It’s definitely true that if you throw more compute at the model, if you make the model bigger, it’ll get better. It’s kind of like it’s the most trustworthy way to improve models. It’s also the dumbest. Right? Like, if all else fails, just make it bigger. And so for folks who have a lot of money, that’s a really compelling strategy. It’s super low risk. You know it’s going to get better. Just scale the model up, pay more money, pay for more compute and go. I believe in it. I just think it’s extremely inefficient. There are much better ways. If you look at the past, let’s say year and a half, I guess by now it would be between ChatGPT coming out, or GPT-4 coming out, and now GPT-4, if it’s true what they say, and it’s 1.7 trillion parameters this big MoE, we have models that are better than that model, that are like 13 billion parameters. And so the scale of change, how quickly that became cheaper, is absurd, kind of surreal. And so, yes, you can achieve that quality of model just by scaling, but you probably shouldn’t.
Harry Stebbings: Do we continue to see that same scaling advantages, or does it actually plateau at some point, as you said there, we always hear about Moore’s Law. At some point, it just becomes a better calculator for the iPhone.
Aidan Gomez: It certainly requires exponential input. You need to continuously be doubling your compute in order to sustain linear gains in intelligence. But I think that probably goes on for a very, very, very long time. It’ll just keep getting smarter. But you run into economic constraints, right? Not a lot of people bought the original GPT-4, certainly not a lot of enterprises, because it was huge. It was massive. Super inefficient to serve. So costly, not smart enough to justify that cost. There’s a lot of pressure on making smaller, more efficient models smarter via data and algorithms methods, rather than just scaling up due to market forces. Just pressure on price.
Harry Stebbings: Will we live in this world of unbundled verticalised models, which are much more efficient and smaller, designed for specific use cases. Or will there be much larger three to five models which kind of rule it all?
Aidan Gomez: There will be both. There will be both. The one pattern I think we’ve seen emerge over the past couple years is that people love prototyping with a generally smart model. They don’t want to prototype with a specific model. They don’t want to spend the time fine tuning a model to make it specifically good at the thing that they care about. What they want to do is just grab an expensive big model prototype with that, prove that it can be done, and then distill that into an efficient focus model at the specific thing they care about. That pattern has really emerged. I think we’ll continuously exist in a world of multiple models, some focused and verticalized, others completely horizontal…
…Aidan Gomez: Yeah, sometimes we do. Sometimes we do. There’s this very obvious next step for models, which is you need to let them think and work through problems. You need to let them fail. They need to try something. Fail, understood why they failed, roll that back, and make another attempt. And so, at present, there’s no notion of problem solving in models.
Harry Stebbings: And when we say problem solving, that is the same as reasoning, correct?
Aidan Gomez: Yeah.
Harry Stebbings: Why is that so hard? And why do we not have any notion of that today?
Aidan Gomez: I think it’s not that reasoning is hard, it’s that there’s not a lot of training data that demonstrates reasoning out on the Internet. The Internet is a lot of the output of a reasoning process. Like, you don’t show your work when you’re writing something on the web. You sort of present your conclusion, present your idea, which is the output of loads of thinking and experience and discussion. So we just lack the training data. It’s just not freely available. You have to build it yourself. And so that’s what companies like Cohere and OpenAI and Anthropic, etc, that’s what we’re doing now, is collecting data that demonstrates human reasoning…
…Harry Stebbings: One thing I’m concerned about bluntly or I look at with hesitation is you see OpenAI price dumping. You see Meta releasing for free and Mark pronouncing the value of open source and open ecosystem. Are we seeing this real diminishing value of these models? And is it a race to the bottom and a race to zero.
Aidan Gomez: I think if you’re only selling models for the next little while, it’s going to be a really tricky game. It won’t be a small market. There will be a lot.
Harry Stebbings: This question may be really stupid. Who’s only selling models and who’s selling models and something else?
Aidan Gomez: I don’t want to name names, but let’s say Cohere right now only sells models. We have an API, and you can access our models through that API. I think that that will change soon. There are going to be changes in the product landscape and what we offer to sort of push not away from that, but to add on to that picture and that product suite. But if you’re only selling models, it’s going to be difficult because it’s going to be like a zero margin business because there’s so much price dumping, people are giving away the model for free. It’ll still be a big business, it’ll still be a pretty high number because people need this tech. It’s growing very, very quickly, but the margins at least now are going to be very, very tight.
And so that’s why there is a lot of excitement at the application layer. And I think that discourse in the market is probably right to point out that value is occurring beneath, like at the chip layer because everyone is spending insane amounts of money on chips to build these models in the first place. And then above at the application layer where you see stuff like ChatGPT, which is charged on a per user basis, $20 a month type thing, that seems to be where at this phase, value is accruing. I think that the model layer is an attractive business in the long term, but in the short term with the status quo, it is a very low margin, commoditized business if we just break it down…
…Aidan Gomez: I think it will be. Right now, chips are just exceptionally high margin and there’s very, very little choice in the market. That’s changing. I think it’s going to change faster than other people think. But I’m very confident.
Harry Stebbings: I think you’ve also seen the stockpiling of GPU’s change a lot. Before there was a sign of real supply chain shortage.
Aidan Gomez: Yes. Yeah.
Harry Stebbings: And now it’s not so much.
Aidan Gomez: No. Yeah. The shortage is going down. I think it’s becoming clear there are going to be more options available and not just on the inference side. Inference is already quite heterogeneous. You actually already have loads of options on the inference side, which is like not the training of the models, but the serving. On the training side, the picture has been, it’s essentially one company that creates the chips that you can use to train big models. That’s still true today. But – actually it’s not true today. There’s two companies. You can definitely train big models on TPUs. Those are actually now a usable platform for super large scale model training. And I think Google has proven that quite convincingly. And then there’s Nvidia. But I think soon, AMD, Tranium, these platforms are going to really be ready for primetime…
…Harry Stebbings: On enterprises, Canva is obviously making a hard push for enterprise. You sell into amazing enterprises. What’s the number one blocker today for why enterprises don’t adopt?
Aidan Gomez: It’s mostly trust in the technology. So security. Everyone is very sketched out by the current state of things. Who’s training.
Harry Stebbings: Sketched out means concerned?
Aidan Gomez: Yeah, yeah, right.
Harry Stebbings: Not like a flop.
Aidan Gomez: Well, they’re hoping that they don’t have a flop. So they’re really scared that someone’s going to take their data, train on it, and put them in some sort of security vulnerability, or that they’ll lose IP. I think that’s a very valid concern because people have been training on user data.
Harry Stebbings: Is there anything you can do to reassure them other than, “hey we’re using new synthetic data?”
Aidan Gomez: Yeah. So our deployment model is set up to do that. We focus on private deployments inside their VPC, on prem. What that means is just, it’s on their hardware, completely privately. We’re not asking them to send data over to us. We’ll process it and give you back the response from the model. We’re saying we’ll bring our models to where your data is. We can’t see any of it.
Harry Stebbings: Will we see the movement back to on-prem in this new world?
Aidan Gomez: When I speak to folks, it’s super conflicted in financial services. Yeah. People are pulling away from cloud. They’re pulling away from cloud. They’re building out their own data center capacity. Everywhere else still seems to be we need to migrate to cloud. It doesn’t make sense for us to have these data centers. I think that it probably depends on the vertical that you’re looking at…
…Harry Stebbings: Are we still in the experimental budgets for enterprise? Everyone’s like, oh, we’re just playing with budgets now. Is that fair? Or are we actually moving into mainstream?
Aidan Gomez: It’s really started to shift. So last year, 100%, it was like the year of the proof of concept. Everyone was sort of testing it out, playing around with it. But recently there’s been a big shift to urgency to get this tech into production. I think a lot of enterprises are scared of being caught flat footed. They’ve spent a year running POCs and testing stuff out. Now they’re sprinting towards, I want to put this into production, transform my product, augment my workforce.
Harry Stebbings: What’s the number one use case for them in terms of what they need or want?
Aidan Gomez: The number one use case…
Harry Stebbings: Because it feels like every board is saying, hey, what’s your AI strategy? And it’s like, what does that actually mean? Is it Klarna, who’s very much, we want to optimize our customer service and we’re going to do that. Is that the number one? Customer service? Is it employee augmentation and productivity?
Aidan Gomez: I think it’s employee augmentation. It’s these models becoming a partner or a colleague to your entire workforce. That’s the most popular use case.
Harry Stebbings: I think Copilot is the right way to do that.
Aidan Gomez: I think Copilot is great and it’s the right idea of augmenting a workforce with an assistant. But it’s siloed again within an ecosystem, so it plugs into Office and the Microsoft suite of products. Enterprises don’t just use Microsoft. They use Microsoft for their email and docs and spreadsheets and then they use Salesforce for their CRM. They have SAP for their ERP, they have some HRM, they have internal software that they built for themselves. And if you really want to augment the workforce, you need to have a platform for developing these assistants, these agents, that’s agnostic to a particular toolset and that prioritizes the tool sets rationally across what people actually use, what the market actually uses. So I don’t think that that’s going to be done by Copilot.
Harry Stebbings: You mentioned the word agent there. Agents is one of the hottest topics in ventureland. Do you think it’s justified, the hype around agent’s agentic behavior, what it does to workflows?
Aidan Gomez: I mean, the hype is justified 100%. That’s the promise of AI. The promise of these models is that they would be able to carry out work by themselves that just dramatically transforms productivity. Once you have a model that can go off and do things independently over a very long time horizon. So no longer like, I’m gonna do this one thing for you immediately in return and I’m done. But like, over the next six months, I’m going to be pumping deals into your top of funnel or something like that, right? Like doing outbound for you. It just completely transforms what an organization can do. The hype is justified. I think my critique would be, is that work going to be most effectively done outside the model builders or within? Who’s going to be best positioned to actually build that product?
Harry Stebbings: Why would it be best done within the models?
Aidan Gomez: Completely depends on the quality of the model. It entirely depends on the model. Like, the model is the reasoner behind the agent, and you have to be able to intervene at that level. If you’re not able to actually transform the model to be better at the thing that you care about. If you’re not the one building the model, if you’re just a consumer of the model, you’re structurally disadvantaged to build that product…
…Aidan Gomez: I think there’s sort of like a meme that’s going around of people saying we plateaued, nothing’s coming, it’s slowing down. I actually really think that’s wrong and not just from like a we need to 10x compute and that type of thing perspective and trust me, it’ll get better. But from a methods perspective. So when I was talking about reasoners and planners and models that can try things, fail and recover from that failure, and carry out tasks that take a long time to accomplish, these are, for the technologists, obvious things that just don’t exist in the technology today. We just haven’t had time to turn our focus there and add that capability into the model. For the past year plus, folks have been focusing on that and it will be ready for production, so we’ll see that come out, and I think that will be a big change in terms of capability…
…Harry Stebbings: What does AI not do today that you think it will do in three years? It will be completely transformative.
Aidan Gomez: I think robotics is like the place where there will be big breakthroughs. The cost needs to come down, but it’s been coming down. And then we need models that are much more robust just because a lot of the barriers have fallen away like before. Reasoners and planners inside of these robots, the software behind them, they were brittle and you had to program each task you wanted it to accomplish. And it was super hard coded to a specific environment. So you have to have a kitchen that is laid out exactly like this.
Harry Stebbings: Exactly the same dimensions, nothing different.
Aidan Gomez: Yeah, so it was very brittle. And on the research side, using foundation models, using language models, they’ve actually come up with much better planners that are more dynamic, that are able to reason more naturally around the world. I know this is already being worked on. There’s like 30 humanoid robotic startups and that type of thing. But soon someone’s going to crack the nut of general purpose humanoid robotics that are cheap and robust. And so that will be a big shift. I don’t know if that comes in the next five years or ten years, it’s going to be somewhere in that range…
…Harry Stebbings: So what have you changed your mind on most in the last 12 months?
Aidan Gomez: The importance of data. I underrated it dramatically. I thought it was just scale. And a lot of proof points have happened internally at Cohere that have just transformed my understanding of what matters in building this technology.
Harry Stebbings: So now it’s the quality of data.
Aidan Gomez: Yeah, quality. Like a single bad example, right, amongst like billions. It’s so sensitive. It is a bit surreal how sensitive the models are to their data. Everyone underrates it.
2. Chip War’s Chris Miller on Putin, China, and The Future – Mario Gabriele and Chris Miller
Which current or historical figure has most impacted your thinking?
Vladimir Putin. He is the most striking embodiment of my belief that you can’t understand people through traditional utility functions.
My background is in Russian studies, and I’m struck by the extent to which our analysis of Putin has changed over time. Twenty years ago, when he first came to power, he portrayed himself – and with some level of accuracy, I think – as a relatively modern leader of Russia. He was reforming the tax system and doing stuff that political leaders do. When we talked about his motivations at the time, the focus was often very financial. I remember very distinguished economists who I respect greatly saying, “Isn’t it the case that Putin is primarily driven by money?” And indeed, there are lots of examples of Putin being hugely corrupt and his friends stealing all sorts of stuff. He’s got his gaudy palaces on the shores of the Black Sea.
But we’ve learned that it’s not all about money. When he invaded Ukraine in 2022, Putin cited Peter the Great and Catherine the Great as justifications for territorial conquest. It’s an illustration that “modern people” are not always driven by modern impulses. The desire for power and glory and control, the desire to be on top and dominate others – for better or worse – are central to many people’s utility functions. These impulses might seem more base, but I think, to some degree, they’re present within all of us. You ignore them at your peril…
…What is the most significant thing you’ve changed your mind about over the past decade?
I’ve changed my mind about the usefulness of thinking like an economist. Even though I may criticize them sometimes, I have great admiration for economists. But they think of everything in terms of utility functions and how to maximize them. They only know how to calculate that in dollars and cents. Though that’s valid, I’ve come to appreciate its limitations.
I’ve spent a lot of time over the past decade studying great entrepreneurs and geopolitical competition. Fundamentally, neither founders nor countries think like economists. Great founders may have shareholders who would like them to consider return on equity, but that’s not how they make decisions. Think of Jensen Huang ten years ago – even though Wall Street was warning him against it, he still poured Nvidia’s money into building out CUDA and the ecosystem around it. If your mode of thinking is purely economic – focused on return on equity or maximizing shareholder value – you miss a lot of what actually drives competitive, successful people.
The same thing is true at the international level. Governments don’t think like economists, either. They try to maximize glory or territory or reputation or power. Like great entrepreneurs, sometimes they simply want to win, just for the sake of besting an adversary. There are ultimately so many things that drive nations and the humans within them that are non-quantifiable. Often, they’re much more significant than strictly quantifiable economic variables…
…What risk are we radically underestimating as a species? What are we overestimating?
We’re underestimating the risk of a great power conflict – World War III. World wars happen roughly every half-century. We shouldn’t forget that. Whether as part of a world war or not, the risk of a nuclear weapon being used in conflict within the next 50 years also seems highly plausible.
You can see points of tension across the border between China and the Western sphere. You see it in the South China Sea with the Philippines, in the East China Sea with Taiwan, and in the Himalayas with India – and those are just the border disputes. It’s easy to imagine how that could spiral in an escalatory manner.
If you put a dollar value on the cost of this kind of conflict, it would be measured in the many trillions. Yet the amount of time we spend thinking about it is not remotely commensurate with that outcome. Some people console themselves by saying, “It’s high magnitude but low risk, so the expected cost is low.” I’m not so sure about that. If you talk about the risk this year, maybe it’s low. But if you think about it over the next decade and factor in the risk compounding every year, suddenly, I don’t think those assumptions hold.
If you think the risk is high, we have two options. You can either offer concessions or build up your capabilities to deter more successfully. From the US perspective, we’ve been doing a little bit of the latter and a little bit of the former under Biden – but not much of either. I think it’s intellectually coherent to say, “Let’s do more of one or more of the other.” I think it’s not intellectually coherent to say, “Let’s just do a little bit of both,” when in reality, defense spending is at historic lows relative to the post-Cold War period.
3. Joel Greenblatt: Value and Special Situation Investment Lecture with Rob Goldstein (Transcript here) – Joel Greenblatt and Rob Goldstein
Rob Goldstein (02:32): We came across Moody’s in early 2000 when it was in the process of being spun off. It was obvious that Moody’s was one of the great businesses that we had ever seen and the problem was it was trading at 21 times forward earnings, and 24 times trailing earnings. So the question we had to ask ourselves was just how much of that greatness was already reflected in the stock price. Just to give a little perspective, typically at that time, we would buy stocks at 10 times earnings and sell it at 14 or maybe even 15 times earnings if we got lucky. So the thought of paying up for a business like this was really a new thing for us. So what I did was I compared Moody’s to Coke…
…Goldstein (04:06): Okay. So several decades ago Buffett figured out that if he identified a really great business he could pay what seemed like a lot of money and still make a fortune. In 1988, Buffett bought $600 million of Coke stock. He paid around 13 times forward earnings, 15 times trailing earnings, and back then the value investment community didn’t understand why that was any great bargain. But 12 years later, the $600 million was worth over $7 billion. So Coke became the classic example of paying up for a great business and making a fortune doing it so that’s why I looked at Coke…
…Goldenstein (05:39): Okay, so there’s three really good things about Coke. [Writes on board: (1) Organic Growth, (2) High ROE, (3) Lasting Competitive ADVANTAGE]. To sum up, those are the three really important things to remember about Coke. In addition it was a relatively easy business to understand and it was a predictable business. Most businesses are neither of those things…
…Okay, my first slide. We have Moody’s historical financials and in the 19 previous years to 2000, revenues had grown at a compounded annual rate of 15% and operating profits have grown at a compounded annual rate of 17%. Not many companies have that kind of terrific performance. In the 19-year period, year-over-year revenue declined only one time and that decline was just a few percent and happened after a period of rapid growth. So you know they’ve done great in the past. But does past success equal future success, and as Moody’s a great business, how should we think about that?…
…Just to explain where this growth came from because it’s important for the rest of the analysis. 30 years ago, when you get a loan, the lending institution would retain that loan. Today, many of these loans are securitized and sold into the capital markets. The guy originating the loan is not necessarily the guy financing the loan. Today there’s trillions of dollars of these securities, including credit card loans, home equity loans, commercial mortgage loans, auto loans, etc. To do these securitizations, you need ratings. Financing loans through the capital markets is more efficient than the old way, so one would expect that the growth would continue. In addition, Europe was way behind the US in terms of their growth curve of issuing these asset-backed securities and Asia was behind Europe. They were just sort of starting to go down that path. So basically there was lots of growth ahead.
We talked about good return on capital which we can get to later. In terms of the lasting competitive advantage, we talked about why there can be no new entrants and we touched on why there won’t be any pricing pressure, because their fees seem reasonable in the larger scope of things. You really have to go to S&P and Moody’s to get ratings, and they both know that, so they’re not going to be very negotiable on price. So the company was in the right place at the right time, and the same factors responsible for the past growth would be expected to continue into the future. So we concluded that Moody’s was a great business…
…This is a price chart of Coke. How much did Berkshire Hathaway make over those 12 years? We’ll assume that he paid $5 a share on $6.88. 12 years later, and his stock was $58 a share – be right around here – and he had collected $4.75 in dividends over that time. Just to keep it simple, let’s assume he was able to earn 6% on those dividends that he received, so let’s value the dividend at $6. So his $5 turned into $64, and he’s got a 23.7% rate of return on his investment, annualised.
I basically pulled these numbers out of an annual report at the time. Question is, why did Buffett do so well on his Coke investment?…
…You’re correct, over the 10-year period, revenues grew at 8.8% and unit case volumes at 7%. Oh it is industry… oh no, the industry’s 4%-5% percent. So over the 10-year period they’ve got some price increases. Of course their cost also went up. They were able to grow their unit case volume to 7% a year, so they had organic growth, they didn’t need that much of it. That translated to 12% operating income growth. There was a little bit of leverage so they got 13% in net income growth and they bought the stock, so they got 15% EPS growth over that time.
The other reason why he did so well was because – we just talked about this – he only had to reinvest 20% of the earnings back into the business. So that meant that in addition to the buybacks he was able to pay our dividends.
Just one formula I’m going to put up on the board because we’re gonna come back to it later, is [writes on board: Growth rate divided by reinvestment rate equals return on equity]. So their growth rate was 12%, reinvestment rate was 0.2, so the return on equity was 60%. So that’s how the business performed and in addition, he did so well because there was big PE expansion. He paid about 15 times forward earnings when he bought the stock and in 2000 when we looked at it, it was trading at north of 30 times expected 12-month earnings.
So how can we expect Moody’s to perform for us over the next 12 years? What growth rate should we assume?…
…Well we settled on 12% and the reason why we settled on 12% is because (1) management guidance was low single digits, and (2) because 12% seemed very reasonable considering the historical operating performance had been so much better in the belief that the same factors responsible for the past growth were going to continue…
…So we felt very comfortable that they could grow at very healthy rates in the future. An estimated 12% operating earnings growth rate for Moody’s happen to be very convenient, because that was Coke’s growth rate during those 10 years we looked at. So for the remaining analysis I could now just focus exclusively on the difference in return on capital and how that impacted the different valuations…
…What would you guess Moody’s return on capital was?
Attendees (33:06): [Indecipherable]
Goldstein (33:09): That’s exactly right. Their return on capital was infinite, because they had no – their $50 million in PP&E, they needed desks and computers for 2,000 employees and that was it. In addition their customers paid on time or in advance. They were in a very strong position. They could demand payment upfront and you typically see that kind of a thing with companies that earn good returns on capital. But the answer was their returns on capital were infinite. Very few businesses like that.
So Coke needed to spend 20% of its earnings on… So they earn a dollar, they spend 20 cents, and you have 80 cents left over. Moody’s would spend nothing. They’d have a dollar left over. So how much more was Moody’s earning stream worth more than Coke? 25%? Okay, a dollar is 25% higher than eighty cents.
Now does this mean that the higher return on capital makes Moody’s worth 25% more than Coke? Well yes and no…
…Goldstein (36:02): The question is, everything else saying the same, does this fact that Moody’s has a higher return on capital mean that their business is worth 25% more than Coke?
Attendees (36:18): Yes, in terms of free cash flow.
Goldstein (36:22): In the short term that’s correct. But in the longer term, they’re not gonna grow at these 12% rate forever. So if you assume that in the very long run that growth rates drop to 5%, then if you go back to this formula [pointing to formula on growth rate, reinvestment rate, and ROE], you see that for Coke, will mean that they need to reinvest 8%-10% of their earnings in the business as their growth rate drops. This formula here is what we use to calculate this 60% ROE for Coke. Growth rate over return on capital equals the reinvestment rate. The growth rate at some point in the future drops to 5%, 20 years down the road or whenever, the return on equity is 60% for Coke we calculated, so that means the reinvestment rate would be 8.3%. The slower the growth, the less capital you need, the more capital you can pay out. So let’s just assume that at some point Coke will be paying out 90%-92% of earnings. So we split the difference instead. Let’s assume that this return on capital thing is going to mean that Moody’s is worth 15% more than Coke. It’s just somewhere in the middle between 25% and 10%, or 8%…
…Okay so you just raised my next point, which was is there something else you need to consider? What are they going to do with the money? So we saw that Coke returns all their excess capital, and we felt that Moody’s was very likely to return all their excess capital. In fact, they were gonna put more of that money into buybacks because that’s what management had said they were gonna do. So we basically took this important point and we could leave it out of our analysis at this point. because they were basically gonna be equal for both companies.
So can we justify 21 times earnings? 13 times 1.15 – the benefits from the higher return on capital – so you can pay 15 times earnings and get the same thing. How about 21 times earnings?
Attendees (46:20): [Indecipherable]
Goldstein (46:31): We concluded that because Moody’s had a much higher return on capital, the business was worth 15% more.
Attendees(46:48): 13 was the PE in Coke in 1988, but you’re saying Moody’s can justify a PE of 15?
Goldstein (46:57): Based on the higher return on capital. We saw that we were going to use similar revenue growth assumptions. Growth rate was the same, ROE was different, and let’s assume this [pointing to reinvestment rate] is the same.
Attendees (47:10): [Indecipherable]
Goldstein (47:18): Yes and I’m gonna get to that in a minute. The analysis I made was, I said what would have happened if back in 1988, Buffett paid 18 times earnings, or $7 a share for his stock. So what would have happened is he still would have done great. He would have made 8 times his money, and he would have had a compounded annual return of 20%. Still a great purchase. So he could have paid 18 times earnings at that point and still have done great. So $5 to $7, increased the price by 40%, gets you your 21 multiple – that’s what we had to get. Which is why we used 1.4. Does that make sense?
Well, let’s say you went back to the 1988 and you said that he couldn’t pay 13 times earnings, he had to pay 18 times earnings, how would he have done on his investment, and he still would have done great. So basically he did so well that he had so much room that he could have paid a lot more for his stock and still had a very good investment. Not as good, but still very good. He would have made 20% a year, each year, over those 12 years, and that 40% number got us to our 21 multiple.
[Equation on board: 13 x 1.15 = 15, 15 x 1.4 = 21]
So we kind of backed into it that way and that was the original analysis. And the reasoning was very sound despite the short cut we used. Actually the first time I spoke in Joel’s class, one of the students like you, said it didn’t – interest rate or something had to do with this – and immediately I knew that interest rates had a lot to do with it. Only I never really thought about it.
So what happened was after Buffett purchased Coke, interest rates over the remaining 12 years dropped from 9% to 6% [uses projector for a chart showing interest rates]. So 9% and over here down to 6%. So if you price the 30-year bonds and said that that 30-year bonds, how would that change in price if rates went from 9% to 6%, the answer would be, it would go up by 42%. If it was a perpetuity, it would go up by 50%, but it’s not, it’s a 30-year bond. So it’d go up by 42% percent and that’s the right way of really looking at things. So, it so happens that – this was somewhat random – but the 42% is basically the 40% that we came up with right here [pointing to the equation on the board of “13 x 1.15 = 15, 15 x 1.4 = 21”].
Attendees (51:50): [Indecipherable] For Moody’s, now interest rates are low…
Goldstein (52:08): Okay, let’s see how Buffett would have done. It’s a very good question actually. Let’s see how he would have done. So if interest rates drop from 9% to 6%, thing’s worth 40% more if rates go up. The way the math works, they’re worth 30% less. So going from 1 to 1.4, it’s 40% up, from 1.4 to 1, it’s 30% less. Had his stock traded for 30% less at the time we did this analysis, he would have had a $40.60 stock, he would have gotten $6 in dividends, he’d have $46.60 over his original $5 investment, he would have made over 8 times his money. I did the math, so I know that’s a compounded annual return of 20%. It’s not as good as the 23.7%, but it’s still very good.
So taking your point. it wasn’t that we were expecting to do 23.7%, we were assuming that if interest rates stayed the same or went down, we could expect to make 20%, and that’s probably what he was looking at when he bought Coke. I don’t think he was betting on lower interest rates although who knows what he was doing. That makes sense right?…
…What happened to Moody’s was a good part and a bad part. The good part is that it did trade up. In October when this stock was actually spun off, it was up 20% from where it was in March. And by that following April – so I guess that had been just over a year – the stock was up 50% from where it had started. Now what happened with Moody’s is – and here’s the sad part because we sold our stock too early on this one – but what happened was in 2001, the business exploded to the upside. Profits didn’t grow 12%, they grew at 40%. In the next year, they didn’t grow at 12%, they grew it 35%. So profits have compounded over the following 6-plus years at 25%, at least 25%. I guess that’s what happens when you use conservative assumptions. But the stock was up 6 or 7 times since then and a lot of those gains came early before earnings really took off.
Attendees (59:10): So do you decide on what price to sell?
Goldstein (59:13): That’s a very good question. We obviously made a bad decision. It went up a bunch, earnings had started to shoot up, yet we thought – we got higher hurdle rates then a guy managing zillions of dollars, so the stock was up 50%, so had to think seriously about selling and putting your money into something else. When you make these analyses, hindsight is 20/20 and everything is so easy in retrospect. But in real time when you’re doing this, you’re obviously worried that stock’s 30 times earnings, what happens if I’m wrong, what happens if things do poorly next year and all of a sudden you’re not paying 30 times earnings, you’re paying 40 times earnings. Now the business looks shaky. So it’s never as easy at the time as it is after the fact. But we sold when it was up 50% or more. of all time.
4. Why China Is Starting a New Trade War – Lingling Wei and Jason Douglas
Interviews with policy advisers in Beijing and people who have consulted with Chinese officials show that China’s leadership faced a pivotal crossroads last year, as the country’s real-estate bust brought the economy to one of its weakest points in decades.
Some advisers argued that China’s economy needed a fundamental rethink, graduating from its traditional heavy reliance on manufacturing and construction and instead prioritizing more domestic consumption—a shift that would make China more like the U.S., and potentially put it on a more stable growth path.
Instead, Chinese leader Xi Jinping ordered officials to double down on the country’s state-led manufacturing model, with billions of dollars in fresh subsidies and credit. He used a slogan to make sure officials got the message: “Establish the new before breaking the old,” or xian li hou po in Chinese.
The “new” in Xi’s model doesn’t mean a pivot to a new growth model. Instead, it is the top leader’s way of refining his idea of what kind of manufacturing for the state to back. In essence, the phrase calls for building industries China wants to dominate for the future—such as EVs, semiconductors and green energy—while also maintaining the country’s traditional areas of strength in “old” sectors such as steel. Any overcapacity problems can be punted to the future…
…Two principles have guided Xi’s thinking, Chinese policy advisers say. The first is that China must build an all-encompassing industrial supply chain that can keep the domestic economy running in the event of severe sanctions by the U.S. and other Western countries. In the top leader’s views, advisers say, industrial security sits at the core of China’s stability as tensions with the developed world rise.
The second is a deep-rooted philosophical objection to U.S.-style consumption, which Xi sees as wasteful.
That leaves China with few options other than investing in exports to stabilize its weakened economy and create jobs to make up for losses in domestic construction…
…Loans to industry, including manufacturing firms, have increased 63% since the end of 2021, while Chinese banks have pulled back sharply on lending to real-estate developers.
Government subsidies, though long central to China’s economic playbook, have also ramped up significantly. Companies listed on the Shenzhen and Shanghai stock exchanges declared $33 billion in government subsidies in 2023, according to figures from data provider Wind—23% more than in 2019…
…In all, 99% of publicly listed Chinese companies now disclose some form of subsidy, according to the Kiel Institute, a German think tank. China spends about 4.9% of its gross domestic product on nurturing industries—several times higher than the U.S., Germany and Japan, according to Scott Kennedy, a China expert at the Center for Strategic and International Studies in Washington.
Craig Allen, president of the U.S.-China Business Council, a lobbying group for American companies in China, said Xi’s manufacturing fixation was on display when he met recently with the governor of one of China’s poorest farm provinces.
When Allen asked the governor about his economic priorities, the governor listed semiconductors, software, biotechnology, robotics, aerospace, batteries, and EVs.
“I would have thought that addressing the immediate needs of his overwhelmingly rural constituents, such as improving agricultural harvests, might be at the top of his economic priorities list,” Allen said.
The fire hose of financial support looks set to keep spraying. The People’s Bank of China in April said it set up a new facility with roughly $70 billion to help bank lending to tech firms. In May, a national fund aimed at financing semiconductor production raised $48 billion from state-owned banks and other government-linked investment vehicles…
…“China’s production of advanced electric vehicles, lithium-ion batteries and photovoltaic products, first met our domestic demand, but also enrich global supply,” Chinese premier Li Qiang said in an address to the World Economic Forum’s June meeting in Dalian, China. The real source of China’s manufacturing edge isn’t government subsidies but its huge scale, which helps pin down costs, he added…
…China has added capacity to produce some 40 million vehicles a year, even though it sells only around 22 million at home. It’s on track to make around 750 gigawatts of solar cells this year, despite only needing 220 gigawatts domestically in 2023. And it is expected to account for 80% of the world’s new supply this year in basic chemicals such as ethylene and propylene, used to make garbage bags, toys and cosmetics—even though prices in China have been falling for 19 months, a sign of oversupply.
At the same time, output of steel, one of China’s “old” industries, increased last year despite waning domestic demand due to the continuing property crisis. Industry executives say Beijing has been prodding them to invest more in upgrading steel production through clean technologies and other means…
…China has suffered from persistent overcapacity in the past, at times raising ire from its trading partners for depressing global prices for steel and other goods.
In 2015, Xi entrusted his economic czar at the time, Liu He, to implement reforms that led to closures of many small and privately owned steel mills and other businesses. For a while, it seemed as if Xi and his economic team were ready to finally tackle overproduction.
But as tensions with the U.S. escalated in recent years, and China’s economy weakened, Xi’s views changed, Chinese policy advisers say. He grew more concerned about ensuring China could produce everything it needed in the event of a conflict with the U.S., and became less sympathetic to Western complaints.
5. What is behind China’s perplexing bond-market intervention? – The Economist
Many governments live in fear of bond-market “vigilantes”, investors who punish errant policies by aggressively selling the sovereign’s debt, driving down its price and thereby pushing up its yield. Financial regulators also worry about bond-market malfunctions, such as unsettled trades, when one party to a transaction fails to honour its promises. These mishaps can send ripples of anxiety through an entire financial system.
Such fears do not seem to apply to China’s financial authorities. On August 9th regulators in the southern province of Jiangxi ordered several rural banks not to settle their recent purchases of government bonds, according to Bloomberg, a news service. Similar lenders elsewhere have also been reported to the People’s Bank of China (PBoC), the country’s central bank, for using their own accounts to buy bonds on behalf of others. Rural banks have been instructed to stick to their main business of lending to local enterprises, rather than to the central government.
The measures are part of an attempt by the central bank to stem a relentless rally in the government’s bonds. Earlier this month yields dropped below 2.1% on ten-year securities, down from almost 2.6% at the start of the year. The causes are clear: China’s economy has slowed, borrowers have retreated and inflation has vanished. Nonetheless, officials have been warning since April that yields would not stay low for ever. In July the PBoC unveiled plans to sell government securities borrowed from other financial institutions if required. The central bank was, in other words, “preparing to short its own government’s bonds”, as Adam Wolfe of Absolute Strategy Research, a consultancy, put it. In the end, the bank left the vigilantism to other members of its posse. On August 5th state-owned banks sold bonds heavily, driving the price down and the yield back up a notch…
…In the long run, the best way to lift yields is to warm up the economy, which is likely to require more borrowing and spending from the central government. Its fiscal stimulus would be more powerful if the central bank supports spending with further interest-rate cuts. In other words, yields may have to fall before they can rise. If China’s government is to succeed in reflating the economy, the PBoC will need to act like an accomplice, not a vigilante.
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 Alphabet (parent of Google), Meta Platforms, Microsoft, and Salesforce. Holdings are subject to change at any time.
