Money Stuff: All Public Companies Are Alike

The thing investors want is uncorrelated returns. What you worry about, as a big institutional investor, is all of your stocks going down at the same time. If you could buy some stuff — private equity or catastrophe bonds or gold — that doesn’t go down when all the stocks go down, that would be good for you, and you would pay a premium for it. This is true within stocks, too: Some stocks don’t go down when all the other stocks do, and those stocks are in theory more valuable. Investors should pay more for those uncorrelated stocks than they’d pay for more typical stocks that are correlated to the broad market. Those uncorrelated companies should have a lower cost of capital: They should be able to sell more stock and raise more money to do more projects, because their projects are in an important theoretical way more desirable. The market wants to fund projects that won’t lose value when everything else loses value.

This is all a bit fragile, though. Stock prices go up and down in part for fundamental business reasons (the company sells more widgets, etc.) and in part for supply-and-demand reasons (investors like the stock so it goes up). A company’s fundamental business might be uncorrelated (or negatively correlated) to the rest of the stock market, but its stock is owned by investors, and those investors can create correlations. If the stock market goes down, the company’s stock might not go down because of its uncorrelated business, but those investors also own other stocks that went down, so they might decide to dump the stock to raise cash, which will make the stock go down. And so the company’s stock will become more correlated with the stock market, not because of its business but because of who owns it. 

So we talked a few weeks ago about the effect of high-frequency trading on the cost of capital of small companies. There are small off-the-beaten-track companies whose stocks do not move much with the broader stock market. This might be because their businesses are genuinely uncorrelated with the broader market, or it might just be that no one is paying attention to them. But when high-frequency electronic trading firms start trading those stocks, they become more correlated with the broader market, because (1) the HFT firms are now paying attention to them and (2) they are paying attention in a somewhat generic way, assuming that they will be correlated with the market — using simple heuristics like “if the stock market goes down we should lower our price for this stock” — and thus causing that correlation. And this raises the cost of capital for those companies: They become more correlated to the market, so investors are less excited about giving them money.

The story here might be something like “as markets get more efficient at processing and disseminating information, they get more correlated.” The good technology of modern markets tends to make all the stocks move on the same news, making them more correlated and raising their cost of capital.

Of course there are offsetting tendencies. Investors want things other than uncorrelated returns. They want liquidity, for instance, and if stocks all become more liquid then that should lower their cost of capital. Or just in general if more money flows into some country or sector or industry, that will lower its cost of capital, even if it also makes it more correlated to the broader stock market. Probably on balance deep liquid global stock markets should lower the cost of raising capital to do business, even if they make companies more correlated.

Here (via Hanno Lustig) is a very cool paper on “The Real Cost of Benchmarking” by Christian Kontz and Sebastian Hanson of Stanford. Intuitively, the most obvious thing that would make a stock correlated with the broader stock market is being in a stock index. If a stock is in the index, then all the index funds will buy it when they get inflows and sell it when they get outflows: It will be owned by generic owners of the stock market, so it will trade like a generic stock. Because it trades like a generic stock, its correlation to the broad market — measured in the standard capital asset pricing model by beta, the sensitivity of the stock’s price to market moves — will go up. And in fact they find that: “Using exogenous variation from Russell and S&P 500 reconstitutions, we show that inclusion in a benchmark stock index increases a stock’s CAPM beta.” [2] As indexing has increased over recent decades, the beta of the average stock has increased: More stocks are more correlated to the broad market. [3]

In the standard capital asset pricing model, a company’s cost of capital should go up when its beta goes up. It is not obvious that this would be true in the case of index inclusions. After all, getting added to an index is generally good for a stock: Index funds control a lot of money, and they buy the stocks in the index. “The S&P 500 Pop Is Back From the Dead,” my Bloomberg Opinion colleague Jonathan Levin wrote this month: Traditionally stocks go up when they get added to the index. When people want to buy a company’s stock, that should lower its cost of capital. 

But there is an important intellectual oddity here. The standard capital asset pricing model is taught in business school, and it says that a company’s cost of capital is a function of (1) the risk-free interest rate, (2) the stock market’s equity risk premium and (3) the company’s beta, that is, its sensitivity to market moves. The model does not say anything about being in the index. At some level everybody knows that stock investors care about stuff other than beta: Index funds only buy stocks that are in the index, ESG investors only buy stocks with good environmental records, meme investors only buy stocks whose executives do interviews with no pants, etc. But that stuff is hard to quantify and model. Whereas you learn the CAPM in business school. Kontz and Hanson write:

The inelastic demand of benchmarked funds for benchmark constituent stocks raises their price, but also increases their comovement. These forces have opposing effects on the discount rate: the increased stock price lowers the implied discount rate and incentivizes investment, while greater comovement increases exposure to market risk and discourages investment. As such, the total effect of benchmarking on discount rates and optimal investment is ambiguous. Second, we assume that managers are boundedly rational and behave exactly as they are taught to in corporate finance textbooks and MBA classrooms: they use the weighted average cost of capital implied by the CAPM to discount cash flows.

The assumption that managers practice what textbooks teach is key to our mechanism. Managers who set discount rates using their stocks’ CAPM beta observe an increase in comovement after benchmark inclusion and infer that their cost of equity has increased. However, the price effect does not enter the CAPM-based discount rate they compute. The failure to fully internalize the effects of benchmarking leads managers to perceive an increase in their cost of capital. Consequently, benchmarking has an unambiguously negative effect on firm investment.

Getting added to the index probably lowers a company’s cost of capital, but to a manager with a spreadsheet it looks like it raises the cost of capital. [4]  They compute that the average mid-sized public company faces an apparent increased cost of capital of about 1 to 2 percentage points, and that this actually drives capital decisions:

Assuming a 6% equity risk premium, the increase in the equal-weighted beta translates into an increase of more than 200 basis points (bps) in the CAPM-implied cost of equity.  Importantly, changes in fundamental risk or leverage do not drive this increase. Instead, [indexed ownership] and beta vary systematically across the market capitalization ranks used in the construction of benchmark indices. For example, the average levels of [indexed ownership] and beta change around the assignment thresholds for the Russell 1000 and 3000. This phenomenon has tangible consequences for corporate policy: we find that firms accounting for over 70% of annual capital expenditures in Compustat experienced an increase in beta. In the cross-section, firms with higher [indexed ownership] invest less and issue less equity, suggesting that the growth of benchmark-linked investing and the institutional design of benchmark indices impact real and financial decisions.

Higher index-linked ownership “results in lower investment and increased payouts to shareholders.” 

It is a strange little story about efficiency creating inefficiency. Index funds make investing cheaper and better and thus increase the supply of capital to public companies. But they also make those companies more similar to each other, which raises their perceived cost of capital, which makes them more reluctant to invest and quicker to return money to shareholders. 

Meanwhile in private markets every company is a special snowflake, investors are happy to fund risky bets, and the companies know that. Public markets make all the companies more alike, so they are reluctant to make bold investments. The action is in the private markets because the public companies are all the same.

Sure okay the president of the United States is getting into the sports gambling business:

Donald Trump’s social-media company is getting into the prediction markets business, where record betting is attracting a host of new entrants.

Trump Media & Technology Group Corp. plans to make prediction contracts available on its Truth Social network, allowing users to bet on events ranging from political elections to inflation-rate changes, according to a statement on Tuesday. Initial testing of the service, called Truth Predict, will start “in the near future,” the release said.

The initiative adds to recent efforts by Trump Media to capitalize on its retail following and comes just after trading volume on the leading marketplaces, Polymarket and Kalshi, hit a fresh record. Investors’ voracious appetite for betting on real-world events has prompted firms like CME Group Inc. and Intercontinental Exchange Inc. to look for ways into the business.

Come on. It is always clarifying to replace the term “prediction markets” with the term “sportsbook,” and terms like “betting on real-world events” with “betting on football.” Here is Trump Media’s announcement:

Truth Social users will be able to trade prediction contracts related to major events and milestones, such as political elections, interest and inflation rate changes, commodity prices on gold and crude oil, events across all major sports leagues, and more using the new product technology called "Truth Predict." Prices will update in real-time, allowing users to react instantly to developments in major current events.

"We are thrilled to become the world's first publicly traded social media platform to offer our users access to prediction markets," said Devin Nunes, Chairman and CEO of Trump Media. "Truth Predict will allow our loyal users to engage in prediction markets with a trusted network while harnessing our social media platform to provide totally unique ways for users to discuss and compare their predictions. With more than $3 billion in financial assets as of the end of the second quarter, and having posted our first quarter of positive operating cash flow after going public just last year, Trump Media is well-positioned to leverage our strong balance sheet and existing social media capabilities to create a new standard
for access to prediction market platforms. For too long, global elites have closely controlled these markets - with Truth Predict, we're democratizing information and empowering everyday Americans to harness the wisdom of the crowd, turning free speech into actionable foresight.”

Global elites! The actual prediction markets are operated by Crypto.com. Here is Crypto.com’s prediction markets web page:

Notice the button that is bigger than the other buttons! 

At Bloomberg Intelligence, Jackson Gutenplan and Geovanni Alvarado have a note out today titled “Sports Betting Dominates Prediction Markets”:

Kalshi's volume surge in September coincided with the start of the NFL season, and our analysis found 69% of September volume was in NFL and college football-related contracts. In total, $2.5 billion was wagered on sports at Kalshi in September, 88% of all volume. Though other markets are also active — $89 million wagered on crypto, $108 million on finance/economy (including Fed decisions) and $61 million on politics — Kalshi is essentially the first national sports-betting ecosystem.

Super Bowl futures can now be in one's brokerage account, and this may expand to prop bets and parlays, erasing any line between trading and gambling. Given an ETF can be structured around derivative futures, if the status quo continues it's conceivable to have sports bets in an IRA. When will we have the first Knicks Championship ETF?

Gutenplan and Alvarado add — as I keep pointing out — that it is not at all clear that this is legal: These contracts are regulated by the US Commodity Futures Trading Commission, whose rules explicitly prohibit event contracts that relate to “gaming.” Everyone, including Kalshi and the CFTC, understood that “gaming” included football bets until about this year. And then Kalshi started offering football bets and the CFTC just quietly let it. There was no new rulemaking or interpretive release or public hearing about how now football bets are not “gaming,” and the CFTC has not repealed its rule prohibiting “gaming” contracts. Everyone is just ignoring it. 

You could imagine that that could change at any time: The CFTC could shut all of this down without any new rulemaking, just by enforcing its existing rules. But it won’t! If the president is running a sportsbook, surely the CFTC won’t stop him.

In other news, you can take Bloomberg's Markets Pulse survey about prediction markets here.

OpenAI was, for a while, a nonprofit organization devoted to building artificial general intelligence for the benefit of humanity. This turned out to be an expensive proposition, so OpenAI needed to raise a lot of money, including from investors who wanted a return on capital. This was a bit hard to square with the nonprofit mission, but they had some ideas. For instance, OpenAI raised a lot of money from Microsoft Corp. by offering it a combination of equity-like investment returns and a commercial partnership in which Microsoft could use OpenAI’s models in its products.

But all of that was just temporary, just stops on the road to artificial general intelligence for the benefit of humanity. It’s not like OpenAI was going to build artificial general intelligence for the benefit of Microsoft. There were assurances against that:

1. Microsoft’s investment returns were capped.
2. Microsoft’s commercial deal would not give it access to OpenAI’s models that achieved artificial general intelligence (AGI): Microsoft was providing interim funding in exchange for interim models, not for control of future AGI.
3. OpenAI was controlled by its nonprofit board, which had no fiduciary duties to its investors. When it achieved AGI, its board might (should) decide to give it away for the benefit of humanity, rather than to hoard it for profit. “It would be wise to view any investment in OpenAI Global, LLC in the spirit of a donation,” it gloriously told investors, “with the understanding that it may be difficult to know what role money will play in a post-AGI world.”

This has all slowly ebbed away, and now OpenAI is running slop video feeds and looking into porn. It has also become a for-profit company:

OpenAI is giving its long-time backer Microsoft Corp. a 27% ownership stake as part of a restructuring plan that took nearly a year to negotiate, removing a major uncertainty for both companies and clearing the path for the ChatGPT maker to become a for-profit business.

Under the revised pact, Microsoft will get a stake in OpenAI worth about $135 billion, the companies said in a statement Tuesday. In addition, Microsoft will have access to the artificial intelligence startup’s technology until 2032, including models that achieved the benchmark of artificial general intelligence (AGI), a more powerful form of AI that most say does not exist yet.

Microsoft will also continue to be entitled to receive 20% of OpenAI’s revenue, according to people familiar with the matter, who spoke on condition of anonymity as the information is not public. But as part of the new pact, OpenAI can pay more later. In a blog post, the companies said a revenue share agreement remains in effect until an expert panel verifies AGI.

With the agreement, OpenAI said its corporate restructure is now complete. The company had spent much of this year working to form a more traditional for-profit company. Microsoft, which backed OpenAI with some $13.75 billion, was the biggest holdout among the ChatGPT maker’s investors, Bloomberg News has reported. …

The restructuring had been under review by the state attorneys general of Delaware and California. In a statement, Delaware State Attorney General Kathy Jennings said her office had decided not to object to the for-profit shift after a long review process in which she and her California counterpart, Rob Bonta, urged OpenAI to give the nonprofit more control over the new for-profit entity.

Here is the joint OpenAI/Microsoft announcement of the deal, which modifies various aspects of the commercial relationship. Starting with:

Once AGI is declared by OpenAI, that declaration will now be verified by an independent expert panel.

Microsoft’s IP rights for both models and products are extended through 2032 and now include models post-AGI, with appropriate safety guardrails.

Microsoft’s IP rights to research, defined as the confidential methods used in the development of models and systems, will remain until either the expert panel verifies AGI or through 2030, whichever is first.

When OpenAI was a nonprofit building AGI for the benefit of humanity, you sort of had to assume its good faith. If it achieved AGI, it would declare that it had achieved AGI; its commercial arrangements would fall away and it would usher in a new era of universal abundance.  Now that OpenAI is more of a, you know, gigantic capital-hungry commercial enterprise, you don’t have to make the same assumptions. If OpenAI declares “hey we achieved AGI so we’re ending our revenue-sharing agreement and cutting you off from our models so we can sell them to someone else,” you can check. You can demand an independent expert review of whether OpenAI has actually achieved AGI. And if it has you can keep selling the products. It may be difficult to know what role money will play in a post-AGI world, but OpenAI and Microsoft have a guess! The guess is “money will still be great, and we will get a lot of it, so we should figure out how to split it.”

Big law firms tend to specialize in certain types of deals, and in representing one or another side of those deals. There are firms that do a lot of sell-side mergers and acquisitions and those that do more buy-sides, firms that represent creditors in bankruptcy and those that represent debtors, firms that represent companies doing initial public offerings and firms that represent underwriters, etc. If you regularly represent one side of a deal, you want to appeal to the people who do that side of the deal: You want to be an aggressive negotiator on their behalf; you want to push contract terms in the direction that is good for them. You want to be tough, because they want tough lawyers on their side. 

But if your client is a repeat player, you can’t be too tough. [5] If you are too mean to the other side, that will make life harder for  your client. The Financial Times has a fun story about Kirkland & Ellis:

Kirkland & Ellis has given its lawyers training on their communication style as it tries to combat a reputation for uncooperative behaviour in negotiations between its private equity clients and their investors.

The efforts follow frustration among investors in private capital funds, known as limited partners, which was vented at an industry event last year.

Organisers of the event in New York asked LPs to name one thing they would like private equity groups to do if they could wave a “magic wand”. Displayed prominently in a word cloud of responses was: “Fire K&E”, according to people who attended.

The global buyout sector is coming under increasing fundraising pressure, making relationships between private capital groups, many of them represented by Kirkland, and their investors even more important. 

The world’s largest law firm by revenue has a prominent practice representing private equity groups in negotiating terms with investors, such as pension funds.

Yes, look, if the other side of a negotiation regularly hates you, that implies that you are doing something right for your client. Some things are zero-sum, and you want to win as many of those as possible. But you can take it too far.

Evercore’s Rainmaker Is Driving PE’s $200 Billion Bright Spot. Wall Street Shrugs Off Credit Worries Even as More Cracks Emerge. HSBC Reviews Ties to Hedge Funds With Credit Fears on the Rise. Nvidia to Invest $1 Billion in Nokia in AI Networking Push. Why Germany’s Merz Is Calling for a Joint European Stock Exchange. Trump Considers Fed Chair Selection by Year-End From Slate of Five Finalists. $100bn stock swings expose ‘fragility’ beneath Wall Street rally. Qualcomm Launches AI Chips to Challenge Nvidia’s Dominance. Texas Sues J&J, Kenvue Over Alleged Autism Risks from Tylenol. Exxon Mobil Sues California Over Looming Climate Disclosure Rules. Forge Global explores sale after share price plunge. “To fire them, we might have to find a concierge to kick down a door in an L.A. hotel and revive them after a three-day cocaine binge. We needed a strong human-resources department.” CEOs Are Furious About Employees Texting in Meetings.

If you'd like to get Money Stuff in handy email form, right in your inbox, please subscribe at this link. Or you can subscribe to Money Stuff and other great Bloomberg newsletters here. Thanks!