A classic conflict of interest occurs when the buyer of something gets reimbursed for part of the cost by someone else. Schematically:

1. Alice is a repeat consumer of a thing, and Bob is a producer of that thing.
2. Carol is someone — Alice’s employer, her insurer, a government agency, etc. — who has an obligation to reimburse Alice, say, 60% of what she pays for the thing. (Or to pay 60% of the cost directly to Bob.)
3. The thing costs $100.
4. Alice pays Bob $100 and gets the thing.
5. Carol pays Alice back $60.
6. Net, Alice pays $40 and Carol pays $60; Bob gets $100.
7. Alice and Bob do this every week, and they get to chatting.
8. Bob asks: “If I raised the price to $200, would Carol pay you $120?” “Yup,” says Alice. “And she wouldn’t push back on the price?” “Nope, she just pays 60% of whatever price I pay; I’m the one who negotiates the price.”
9. “Why don’t I just raise the price to $200 then,” says Bob. “Well, because I’d be out of pocket $80, and I can’t afford that, so I’d stop buying from you,” says Alice.
10. “Hmm,” says Bob. “What about this. What if I raise the price to $200, you give me $200, Carol reimburses you $120, and then when her check clears I give you a $70 refund? That way I get $130 instead of $100, and you pay $10 instead of $40. We are both $30 better off. And Carol is $60 worse off but, and this is really the critical point, she isn’t here.”
11. “Yes this is good,” says Alice. “The only problem is that if you give me a $70 refund, Carol might notice that the net price is only $130. Can we call it something other than a refund? Like if I pay you $200 for the thing, and Carol reimburses me $120, and then you write me a $70 check as a Christmas present, Carol won’t have anything to complain about.”
12. “Sure,” says Bob, and they do that. The new price of the thing is $200; Alice pays $200; Carol reimburses $120; Bob gives Alice a $70 Christmas present. Net, Bob gets $130 ($200 minus $70), Alice pays $10 ($200 minus $120 minus $70), and Carol pays $200.
13. “Hey I’ve been thinking,” Bob says after a while. “What about $300? Carol would reimburse $180, I could give you a $150 Christmas present, I’d be better off than I am now, and you’d actually make money each time you buy the thing.”
14. “What about $500,” says Alice.

There are all sorts of versions of this, [1] many of them in the US healthcare space, where (1) no one has any intuitions about the correct price of the thing [2] and (2) most consumers are getting reimbursement from someone. Alice is a patient, Bob is a drug company, Carol is Alice’s insurer, Carol pays 80% of the cost of Alice’s drugs (that is, Alice has a 20% copay), and instead of a “refund” or “Christmas present” Bob gives Alice a “charitable copay assistance program.” Bob can charge $10,000 for the drug instead of $100, and Alice is indifferent, since Bob — or Bob’s affiliated charity — pays her copay. Bob is not indifferent, since he gets $8,000 (from Carol) instead of $100.

In the health insurance space, this stuff is pretty well known, but that doesn’t mean it is prohibited, exactly. It is regulated. There is some social value in Alice getting cheap access to expensive medicines; drug companies can help with copay assistance programs as long as they don’t do it in ways that are too nakedly aimed at boosting the prices they charge insurers. Sometimes they get in trouble for going too far.

Or there is the Medicaid provider tax, in which Alice (the buyer) is a US state government, Bob (the seller) is a hospital or other medical provider, Carol (the reimburser) is the US federal government, and instead of a “refund” or “Christmas present” the hospital pays the state an extra tax. The New York Times explains the trade:

In its simplest form, the tax maneuver works like this: When a Medicaid patient goes to the hospital, the federal government and state usually share the costs. The ratio varies from one state to another, depending on how poor the state is, but the federal government often pays around 60 percent of the bill.

States that use provider taxes to get more money usually start by paying the hospitals more. If the federal government is paying 60 percent and the state 40 percent, when a state bumps a payment to $1,030 from $1,000, the federal government chips in $618 instead of $600.

And then the state imposes, say, a $25 tax on the hospital. Net, the state pays $387 (the $1,030 payment minus $618 of federal reimbursement and $25 of tax) instead of $400, and the hospital gets $1,005 instead of $1,000. The federal government has chipped in an extra $18, and the state and the hospital have divided it up between them. 

You can view this narrowly, as a trade between the state and the hospital. From that viewpoint it’s a great trade, an arbitrage, maybe a scam: Both sides get free money by putting one over on the federal government. “For years,” notes the Times, “the use of provider taxes in New Hampshire was openly described as ‘Mediscam’ by state officials.”

Or you can view it a bit more broadly. The federal government, after all, is aware of this. Forty-nine states use some version of it. There are extensive federal rules about how reimbursements work, rules that specify what sorts of taxes do and do not meet the requirements. [3] (Literal refunds of state payments don’t work, but broad taxes on medical providers do. [4] ) The broader view is something like “the federal government wants Medicaid recipients to get treatment, and is willing to pay a bit more than the published reimbursement rates to make sure that happens; for boring institutional reasons the federal government will only reimburse states for 60% of the stated cost of Medicaid services, but it’s fine to inflate that stated cost a bit so that the federal government reimburses, say, 65% of the actual cost.” 

Obviously on the second view, the much more efficient approach would be to outlaw the shenanigans and have the federal government reimburse states for more of the cost of Medicaid. In the limit, if the federal government is going to pay most of the cost of Medicaid, it might be more efficient for the federal government to pay 100% of the cost directly to hospitals, rather than putting states in the middle and creating incentives to game the costs. But, you know, those boring institutional reasons are real, and it’s possible that allowing states and hospitals to team up to gently scam the federal government really is the most practical way to get the government to pay its socially optimal share of the cost of medical care.

Or not. The Times also reports:

Long after these taxes have become entrenched, congressional Republicans are now considering curtailing or ending them as one way to achieve the steep federal spending reductions proposed in the House budget. If they did, it would save the federal government about $600 billion over the next decade, a large chunk of the $880 billion in cuts that the House committee that oversees Medicaid has been charged with finding. …

The proposals circulating on Capitol Hill to ban the taxes do not include provisions to replace any lost dollars with new funding sources, which would leave some states with big holes in their Medicaid budgets. In some places, more than a third of federal Medicaid spending would vanish. ...

“If you take away this money, that’s a policy decision,” said Robin Rudowitz, director of the Medicaid program at KFF, a health research group. “It’s not an issue of cracking down on fraud.” …

 A 2020 report from the Government Accountability Office estimated that states, on average, were using the taxes to get the federal government to pay an extra 5 percent of their Medicaid bills. 

That is: The current deal is that the federal government pays 65% of states’ Medicaid costs (on average, with wide variation), but for boring institutional reasons that deal is expressed as “the federal government pays 60% of the cost but you can cheat a bit on the costs to get to 65%.” A revenue-neutral way to reform that might be to say “the federal government pays 65%, but no more cheating.” Or you could leave it at 60% and ban the cheating, but at this point, after the cheating has been accepted for decades, that is also a policy decision to cut Medicaid.

I used to be a corporate derivatives structurer at an investment bank, which means that I learned a little bit about options math and dynamic hedging and accounting, and a lot about storytelling. The business of derivatives sales, like any sales business, is about telling an emotionally resonant story showing how your product solves the customer’s problems. If you know her fondest hope, you try to sell her a product that (seems like it) can make it come true. If you know her deepest fear, you try to give her a product that (seems like it) can guard against it. If your products are, you know, stock options, they are not necessarily a natural fit for anyone’s fondest hopes or deepest fears, but you’re paid for creativity. You think of some exciting stories about stock prices and sell the customer upside; you think of some scary stories about stock prices and sell the customer downside protection. The products don’t change much — you can tinker with the parameters, but basically they’re stock options — but the story you tell about them does.

Similarly, if you are in the business of financing trade, President Donald Trump’s tariffs have some obvious downsides for you. If there are big tariffs on all goods coming into the US, trade will probably go down, and you will finance less of it. On the other hand, the tariffs are new, so there are new stories for you to tell. “We finance trade, we could finance the tariffs, that’s something,” is probably the thought process that a lot of trade financing bankers are having. Like this one:

HSBC Holdings Plc is offering a new loan product to US companies struggling to cover the cost of President Donald Trump’s tariffs that have roiled international supply chains.

The London-headquartered bank said on Wednesday its TradePay platform was being extended to directly cover the cost of tariff payments, allowing importers to effectively borrow to meet the increased expenses involved in shipping products into the US.

“By settling import duties directly and frictionlessly through HSBC TradePay, our US clients have more visibility and control over their working capital,” said Vivek Ramachandran, head of global trade solutions at HSBC.

Under the new loans, customers’ import payments will be automatically paid through pre-agreed credit with brokers or a direct deduction using automated clearing house credits, meaning companies will be better able to manage their cash flow and settle duties more efficiently.

HSBC is the world’s largest trade bank and the biggest international bank in China, giving it a crucial role in oiling the wheels of international trade, particularly between the two biggest economies. The US has imposed tariffs as high as 145% on Chinese goods, while China has hit back with retaliatory rates of 125%. Talks aimed at de-escalating the situation are due to take place this week.

Yes, right, “we lend you money to finance your imports, and the cost of your imports is going up so we will lend you more money” is not exactly a new loan product, but it is certainly a timely and topical re-description of an old product. And “the cost of your products has doubled, but at least you can borrow money to pay the extra cost” is not an appealing story per se, but the bankers work with what they’ve got.

I wrote yesterday about OpenAI’s corporate structure. Basically OpenAI is a company that

• is controlled by a nonprofit organization, and
• has raised money from investors (and incentivized employees) by giving them complicated stakes in the economics of its business, stakes with forms like “X% of our profits up to $M, and then Y% of profits up to $N, and then Z% of profits up to $P, but nothing above that.” 

And now OpenAI is planning to convert to a more normal structure, in which

• all those weird economic stakes will be converted into normal stock, so instead of a complicated profit waterfall each investor just gets some percentage ownership of the company “in an amount supported by independent financial advisors,” and
• the nonprofit organization gets the rest of the shares — if all those profit waterfalls work out to the investors getting 71% of the company, the nonprofit gets the other 29% — but its shares are super-voting, so the nonprofit keeps control of the company.

This is an approximate description and abstracts away from some of the mechanics. Those mechanics are particularly complicated with Microsoft Corp. Microsoft is a big investor in OpenAI, and has some complicated series of profit interests that will convert into stock. But it didn’t just hand OpenAI a bunch of cash for those interests. For one thing, it seems to have paid mostly in the form of computing power, not cash. And it has a commercial relationship with OpenAI, with exclusive deals to use OpenAI’s models in its products and cloud services.

And so when OpenAI goes through its process of slimming down and simplifying its capital structure, converting weird profit-sharing deals into shares of stock, I suppose the question might come up: What counts as capital structure? What bits of Microsoft’s relationship with OpenAI consists of economic quasi-ownership that should get converted into stock, and what bits consist of commercial deals that should just continue? Is “20% of revenue, plus 75% of the first $17 billion of profits, plus 49% of the next $83 billion of profits” convertible into “___% of stock,” or into “well you keep getting the 20% of revenue and you also get ___% of stock,” or something else?  

The Information reports:

As OpenAI moves ahead with a corporate restructuring to appease its investors as well as regulators, it needs to get clearance from Microsoft, its biggest outside shareholder and business partner.

One part of OpenAI’s plan involves reducing the percentage of revenue it shares with Microsoft. In financial projections OpenAI shared with investors, that percentage would drop by at least half by the end of this decade. Microsoft, whose servers power OpenAI products, still hasn’t signed off on OpenAI’s desired changes, according to a person who spoke with a senior Microsoft executive involved in the negotiations.

In an existing deal, OpenAI agreed to share 20% of its revenue with Microsoft through 2030—a year in which OpenAI has projected it would generate $174 billion in revenue. In recent weeks, though, OpenAI told some potential and current investors that by 2030, it would only share 10% of revenues with commercial partners including Microsoft, according to private documents. ...

OpenAI and Microsoft have been wrangling for months over amendments to their existing agreement, including changing the 20% revenue-share figure and how long Microsoft will retain the right to use OpenAI’s intellectual property. …

[OpenAI’s restructuring] would also give investors including Microsoft traditional stock instead of shares that entitle them to potential future profits. That seemed good enough for some investors including SoftBank, which reaffirmed its commitment to invest $30 billion in OpenAI despite the nonprofit retaining control over the company, CEO Sam Altman told reporters on Monday.

OpenAI would like to simplify its financial and commercial arrangements, but there’s only so far it can go.

My crude model of President Donald Trump’s tariffs is that the Trump team thinks that too much US economic activity is in finance, and not enough is in manufacturing. After Trump’s “Liberation Day” tariff announcement last month, I wrote that “a world in which the US gives people finance and gets back inexpensive goods strikes me as good for the US. We give them entries in computer databases, they give us back food and clothing: That is a magical deal for us!” But the Trump economic team appears to disagree. They have argued that the US dollar’s status as the world’s reserve currency is bad for us: Foreigners are too happy to buy US financial assets, which drives up the value of the dollar, makes our exports less competitive, and causes a US capital-account surplus that leads to a current-account deficit. Foreigners trade their manufactured goods for our financial assets, and Trump wants them to trade their manufactured goods for our manufactured goods. Presumably, in the desired end state, US factory employment would go up and financial-industry employment would go down.

Now I should say that this is only one possible model of what Trump’s advisers are up to, and they do tend to say lots of contradictory things, so many other models are roughly equally plausible. (My model is perhaps best supported by the statements of Stephen Miran, the chair of Trump’s Council of Economic Advisers, but even he explained his own view of Trump’s economic policy by saying that “I expect policy to be dollar-positive before it becomes dollar negative,” so you could take that however you want. [5] )

But it has been striking to me that the execution of Trump’s tariff policy seems to be broadly a lot of fun for the financial sector and less good for manufacturing. If tariff policy is coy and uncertain and changes every day, that creates enormous volatility. A lot of financial firms make their money, more or less, on volatility: The more asset prices bounce around, the more trading activity there is, and the more money you can make (or lose) by providing liquidity. Manufacturers looking to build factories, though, do not make their money on volatility. They want predictability. If you are going to spend a billion dollars building a factory in the US to take advantage of high tariffs on the rest of the world, you need to have some sense of what those tariffs will be when you open the factory. You have no idea.

So it’s hard to invest in a factory. You know what you can do with your money, while you wait for certainty? You can optimize your capital structure. The Financial Times reports:

US companies are planning to buy back a record $500bn of their own shares, as they seek to deploy their huge cash piles at a time when President Donald Trump’s policies are adding to uncertainty over making capital investments. ...

Trump’s unconventional trade manoeuvrings have in a number of cases made planning for the future more difficult, forcing groups such as Colgate, General Motors and Delta Air Lines to cut sections of their earnings guidance.

Rising buybacks are “a reflection of the fact that global tariff uncertainty is getting in the way of planning for operational investment”, said a former co-head of equity capital markets at a large US investment bank.

My naive view of the Trump industrial policy is that the administration would prefer for companies to spend more money on operational investment and less on stock buybacks, but the effect might be the opposite.

Lawyers Are Quoting $1 Million in Fees to Get Pardons to Trump. Top Trump Crypto Buyers Vying for Dinner Seats Are Likely Foreign, Data Shows. How dealmaking fever hit Italy’s banking sector. Disgraced Ex-Goldman Partner Sought by Malaysia for Extradition in 1MDB Scandal. Milken Private-Credit Crowd Spots ‘Golden’ Moment Amid Chaos. SMBC Teams With Private Lenders to Offer $1.7 Billion of Credit. The High-School Juniors With $70,000-a-Year Job Offers. Geely Plans Bid to Take Zeekr Private at Value of $6.4 Billion.

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!