Money Stuff: The ETF Dividend Flip

Let’s say you own stock in a company that has announced that it will pay a $2 per share dividend to anyone who holds the stock on Wednesday. On Tuesday, the stock closes at $100 per share. If you buy the stock on Tuesday, you receive it on Wednesday — this is called “T+1 settlement” — so you are a shareholder as of Wednesday and get the dividend. If you buy the stock on Wednesday, though, you don’t actually get the stock until Thursday, so you miss the dividend. (Wednesday is the “ex date,” the first day on which you don’t get the dividend if you buy the stock. [1] ) In theory, ceteris paribus, the stock should trade at $98 on Wednesday to reflect the fact that you don’t get the dividend. The stock’s price on Tuesday represents (1) the value of the stock plus (2) the $2 dividend; the price on Wednesday represents (1) roughly the same value for the same stock plus (2) no dividend. So $100 on Tuesday, $98 on Wednesday. Ceteris paribus.

Many countries tax dividends in ways that are bad for foreign investors. In the US, for instance, many foreign investors face a 30% withholding tax on dividend income. And so there is a large financial-industry subculture of finding ways to transform “dividends” into something else that gets better tax treatment. Many of these trades are done with swaps or other derivatives, but here is a very simple one:

1. You are a foreign investor, you own a share of stock, it’s going to pay a $2 dividend, you do not want to “receive” the “dividend” for tax reasons, but you don’t want to give up the $2 either.
2. On Tuesday at 11:59 p.m., just before the dividend ex date, you sell your stock for $100.
3. On Wednesday at 12:01 a.m., you buy it back for $98, and keep the other $2.
4. Now you have the same share of stock that you had before, plus $2.
5. Your $2 is not, however, a dividend. It’s just the result of selling stock at one price and buying it back at another. It’s just capital gains, which are not subject to withholding tax.

That’s pretty good: You sell the stock just before it pays the dividend, and buy it back just after. In theory, ceteris paribus, the stock will go down by $2 — exactly the amount of the dividend — in the two minutes between when you sell and when you buy it back, so your sale price will exceed your repurchase price by exactly the amount of the dividend. You will cash out the amount of the dividend, but it will not be a dividend.

This trade is not quite practical, though, because you can’t really sell your stock at 11:59 (with the dividend) and buy it back at 12:01 (without the dividend). Liquidity is bad at midnight, and even 24-hour-ish stock markets take little breaks to update their computers, so this trade might be difficult to execute. More important, stock trades in the overnight session do not exactly settle T+1, and if you wait until 11:59 p.m. to sell you will still get the dividend. [2]  

So the realistic version of this trade is more like “on Tuesday at 4 p.m., you sell for $100, and on Wednesday at 9:30 a.m., you buy it back at $98.” Here, still, ceteris paribus, you will be cashing out the amount of the dividend, but ceteris are less likely to be paribus. A lot happens overnight! In fact, famously, approximately 100% of stock-market gains occur overnight. Therefore, if you sell your stock on Tuesday at $100 per share, it will probably open on Wednesday at, like, $98.50 or $99 per share. (Empirically, stocks normally do not drop by the full amount of the dividend on the ex date, in part because stocks mostly go up overnight and in part, probably, for dividend tax reasons.) You will not keep the full $2 for yourself. You will miss out on some of the value of the dividend, because you will not hold the stock for 17.5 hours.

For S&P 500 exchange-traded funds, though, there’s a much better trade, Bloomberg’s Zachary Mider and Denitsa Tsekova report:

Investors hopscotch between BlackRock’s iShares Core S&P 500 ETF (ticker IVV) and Vanguard’s S&P 500 ETF (VOO) so they hold neither fund on the date that it confers a right to receive a dividend. Thanks to the way stock prices move around distribution dates, the investors don’t miss out on any of the return from the dividend. Instead, they effectively take the value of the payout in price appreciation — which isn’t taxable for foreigners. …

Flipping from IVV to VOO and back probably saved foreign investors about $147 million in US taxes last year, according to calculations based on the dollar value shifting between the two funds each quarter. …

The switching trade relies on the way ETFs handle dividends from the stocks they hold. The members of the S&P 500 pay cash dividends to shareholders on various dates. Index funds accumulate that over time and pay it to investors once a quarter. The S&P 500’s annual dividend yield is just over 1% of its market value.

To avoid receiving a fund’s dividend, investors need to have sold their position by market close on the day before the ex-dividend date — the point for each ETF when its shares begin trading without the right to receive the upcoming payment. Since 2024, BlackRock’s ex-dividend dates have been consistently earlier than those of Vanguard, making it easy to toggle between the two providers.

That is: You hold IVV until a day or two before its ex-dividend date. Then you sell it, like, mid-day on Tuesday and simultaneously buy VOO, so that you have no slippage: You hold S&P-500-ETF-in-some-form continuously. [3]  The IVV ex-date arrives, and IVV starts trading lower than VOO: For a few days, VOO trades with a dividend and IVV trades without, so VOO trades higher than IVV by about the amount of the dividend. So you sell your VOO (high) and instantaneously buy back IVV (low), again continuously holding S&P 500 exposure with no slippage. When you sell your VOO, you get a higher price than you pay to buy back the IVV. Which is a dividend, economically, but not for tax purposes.

Mider and Tsekova write:

The switching trade adds to the growing list of techniques that Wall Street and its major investors are deploying to minimize tax bills. While many such tactics rely on a special loophole that helps ETFs defer capital gains, this one is simpler, depending merely on the existence of multiple large, liquid, nearly identical funds. … The switching trades are unrelated to the so-called heartbeat transactions that play a key role in the ETF industry.

We have talked a few times about the fact that an ETF is approximately an investment vehicle that doesn’t pay taxes, and that this technology has been so broadly generalized that maybe one day no investors will have to pay taxes. But this isn’t that; this has nothing to do with that. This is just trading around ex dates. This is just: If you can sell a share of stock before it pays a dividend, and buy an otherwise identical share of stock that pays the same dividend on a different date, you never need to pay dividend taxes.

The economic model of law firms is something like this:

1. Clients want to hire certain lawyers — partners at big law firms — for their judgment, skill and experience.
2. The top partners at top law firms can make $20 million a year or more.
3. If you want to hire a top partner at a top law firm to work on your bet-the-company deal or litigation, you have to pay the market rate for her judgment, skill and experience.
4. But, for reasons of history and tradition, you don’t quite do that.
5. Instead, traditionally, lawyers bill by the hour. You pay an hourly rate for her time, which might be as much as $4,000 for a partner at a big firm.
6. If you divide $20 million by $4,000 you get 5,000 hours a year, or about 100 hours a week. The top partners at top law firms do not bill their clients 100 hours a week. [4]  That would be tiring.
7. The price, to a client, of hiring a top lawyer is not really her stated hourly rate times the number of hours she works on your legal problems.
8. Instead, the price, to the client, is some multiple of her stated hourly rate.
9. The way the multiple gets computed is with associates. Instead of buying 10 hours of the senior partner’s time for $40,000, you buy 10 hours of her time plus 200 hours of time from the associates — more junior lawyers — who work for her and help her analyze your legal problem, review your documents, etc.
10. The associates bill out at like, $1,000 an hour or whatever.
11. In total you pay $240,000 for 10 hours of the partner’s time plus 200 hours of the associates’ time.
12. Out of that $240,000, the partner pays the associates some money but keeps most of it for herself, putting her on track to earn $20 million a year. 
13. For your $240,000, you get 10 focused hours’ worth of judgment, skill and experience from a top lawyer at a top law firm.
14. Plus those associates are definitely reviewing some documents!

This is exaggerated, of course; in fact the law firm associates do lots of good and useful work for their clients. (I was one, once, and I did!) But the basic point is that law firms are tied to the billable hour, and the senior partners’ time is underpriced, both in the sense that clients would (and do!) pay more for it than their hourly rates, and in the sense that the senior partners take home a lot more than their hourly rates times the number of hours they work. Whereas the associates’ time is overpriced, both in the sense that the clients wouldn’t pay the associates’ hourly rates just for the associates’ work, and in the sense that the associates take home a lot less than their hourly rates times the number of hours they work. Billing out the associates’ time is a way to capture more value for the partners.

Now there is artificial intelligence, which can, perhaps, do a lot of the work — document review, drafting, legal research, etc. — that associates do. You could just about imagine a law firm of only senior partners, with the associates’ work done by AI. But how would that firm bill? How would the partners make $20 million a year? You could imagine some answers:

• The partners could bill $25,000 per hour. “AI has made us more efficient, so our rates have gone up.”
• The firm could give its AI agents cutesy names and bill them out as associates. “Your bill is $40,000 for 10 hours of partner time, plus $200,000 for 200 hour-equivalents of token use by Legalina, our AI system.”
• Flat fees. “I’ll apply my judgment, skill and experience, plus some AI, to your legal problem for $250,000.”

Those answers all assume that the market rate for the senior lawyers’ judgment, skill and experience is $20 million a year, and that the current system of associates and billable hours is a convoluted but functional system for paying them their actual market value. Perhaps that assumption is wrong, or will be rendered wrong by AI. Some other answers might be:

• The clients type their legal questions into ChatGPT, get perfectly serviceable answers and pay their lawyers $0. [5]
• The clients say “I was paying you $240,000 for a combination of your judgment/skill/experience and your associates’ grunt work, but now you can do the grunt work cheaply with AI so I’ll only pay you $100,000 for that combination.”

Etc. Here’s a Financial Times story about the problem:

Wall Street banks are pushing large law firms to cut fees, arguing that the business model that has enriched top lawyers for decades is not sustainable in an era of AI. ...

The pressure from some of Big Law’s most valued clients could help overturn the financial model at the heart of the legal industry. Under the so-called leverage model, firms maximise profits for equity partners by billing hourly for work done by large numbers of more junior lawyers, often working late into the night at rates that far outstrip the cost of their salaries.

Much of that work, such as research, document review, assessing contracts and trawling through litigation discovery, can now be done far more quickly using AI. … 

Top lawyers have “for a long time been compensated on the foundation of [associates billing for long hours],” Eric Grossman, Morgan Stanley’s general counsel, told the FT. “Their compensation model is now extraordinarily unstable.”

The ability to complete tasks more quickly could mark “a fundamental altering of the revenue foundation for these mega firms”, he said.

Grossman said the bank was willing to continue to pay large sums for the judgment and talent of the best lawyers, but that by the end of this year most external legal work would be tendered through competitive bidding processes and paid for using alternative arrangements such as fixed fees. 

One possibility is that AI will give the top lawyers more leverage: Instead of working on 20 matters with 20 different associates and making $20 million a year, they can work on 200 matters with 200 different AI instances and make $200 million a year. But probably there will be some pricing pressure.

We talked on Monday about hedging KPIs. My basic points were:

1. Companies (and college football teams) want their executives (and coaches) to do good stuff like increase revenues and profit margins (and win football games).
2. Companies (and college football teams) therefore offer executives (and coaches) contingent compensation packages that pay them big bonuses if they hit certain key performance indicators (like revenue of $X, or profit margins of Y%, or winning Z football games).
3. In a narrow sense, this exposes the companies (or teams) to financial risk: Paying the bonus costs more than not paying the bonus. Perhaps the companies (or teams) should hedge this risk.
4. But not really: You pay the bonus because achieving the KPI is good, for the company (or the football team). Bringing in more revenue or having a higher profit margin is correlated with increasing shareholder value. Winning more football games is also correlated with good economic outcomes for the university: If you win more games, you will recruit more paying students and get more alumni donations and sell more merch and so forth.
5. Therefore it’s just weird to hedge the bonuses, you know? Hedging the bonuses — betting on your company or team to achieve some KPI — is not a hedge; it’s doubling down on your essential business risk. 
6. Nonetheless it appears, from trading on Kalshi, that at least some college football teams hedge their coaches’ bonuses for winning.

Several readers pointed me to a possible explanation, at least of the football thing. When a company pays an executive a bonus for hitting a KPI, the company (1) gets the benefit of the KPI and (2) pays the bonus. College football financing is weirder, though, and it is approximately true that:

• When a college football team wins the national championship, the school benefits (higher enrollment, TV contracts, merch, donations, etc.); but
• When a college football team wins the national championship, a consortium of rich alumni boosters might be on the hook to pay some or all of the coach’s bonus.

From the school’s perspective, this is incentive-aligned and there is no need to hedge. From the boosters’ perspective, though, paying a $3 million bonus really does cost more than not paying it. They might be thrilled to pay it — they got into the boosting business because they want to win a national championship — but it still might hurt a little, and they don’t really share in the school’s extra revenue. Perhaps they should hedge, and perhaps they do. [6]

Back in the glory days of financial blogs, there was one called Long or Short Capital. In 2006, it published a post under the byline Johnny Debacle, with the title “Four Simple Steps to Becoming a Billionaire.” I still think about it from time to time. Here are the steps, in Debacle’s words:

1. Take outrageous risks with extremely high upside.

2. Be the 1 out of 500 million for whom it pans out. This one is key so focus on it.

3. Attribute your wealth creation to your own hard work, your own genius and the power of your business plan. Be sure to stress how your wealth was singularly made possible by your unique endowment of elbow grease, street smarts, common sense, all of which your competitors obviously lacked (proven by how poor they are compared to you).

4. Buy a mega yacht and/or athletic team.

Yesterday Victor Haghani, James White and Jeffrey Rosenbluth of Elm Wealth sort of formalized this approach in a post titled “Do You Really Want to Be a Billionaire?”:

Say Billy expects to earn about $10 million after tax over his career, and save 20% of it, or $2 million. Let’s say he has 50 years to turn that $2 million into $1 billion, a 500x increase.

Now suppose his investment opportunity set looks like this: he can flip a coin 50 times, and each flip has a 60% chance of landing on heads. On each flip, he can bet whatever fraction of his wealth he chooses. We’ll explain later why this is a reasonable, even generous, stand-in for real investment choices.

What’s the best strategy, and what probability of success does it give him? It’s a tricky problem, but it was actually solved back in 1961 in a somewhat obscure paper by UC Berkeley professor Leo Breiman. The answer: with the optimal strategy, Billy can achieve a 7% chance of becoming a billionaire.

Seven percent sounds pretty good. Only about 0.001% of Americans are billionaires, and probably only around 0.05% of people who’ve managed to save $2 million ever get there. A 7% shot is a huge improvement.

But look at the other 93% of outcomes. In those cases, Billy doesn’t just fall short of a billion – he goes broke. That’s because the strategy that maximizes his odds of hitting $1 billion requires him to routinely bet 100% of his wealth, with also means he’ll routinely go broke. Every path that doesn’t end in a billion ends instead with an all-in bet that loses. There’s no soft landing; there’s no “I’m almost a billionaire.”

This is, like the Long or Short Capital post 20 years ago, obviously a stylized depiction of the problem. Nobody is actually offering you coin flips; you need to supply elbow grease, street smarts etc. to find and execute the coin flips. 

Still it does seem to capture something about at least one path to billions, one that comes up sometimes around here. A few years ago we spent a fair amount of time talking about Sam Bankman-Fried’s approach to positive-expected-value coin flips. Literally: When he was an intern at Jane Street, he went around betting on literal flips of literal coins in ways that were perhaps revealing. But also figuratively: Bankman-Fried did become a billionaire by taking outrageous risks with extremely high upside, but then he kept flipping metaphorical coins and oops. “Everything about Sam Bankman-Fried’s life was perfectly optimized for becoming a famous billionaire and an infamous criminal defendant, in that order,” I wrote.

And just this week, we talked about Bankman-Fried’s and his friends’ investments in Anthropic. I wrote that “Anthropic seems to have raised a lot of money specifically from [Effective Altruism] types who love (1) worrying about AI safety and (2) otherwise taking every positive-expected-value bet that comes their way. … For a while everyone was just out there flipping coins for vast wealth or ruin.” That approach definitely appeals to some people. And works for some of them.

Chevron to Invest $7 Billion Doubling Its Venezuelan Output. Volkswagen Fights Chinese Competition — and Its Own Board — in Battle to Survive. S&P Weighs Multibillion-Dollar Spinout of Data Platform Capital IQ. Nvidia Nears $14 Billion Hugging Face Deal This Week. The Rapid Expansion That Built Bally’s Is Now Under Pressure. Uber to Cut 10% of Jobs, or 3,300 Roles, to Slash Bureaucracy. College M&A Tool Helps Schools Connect in Grim Higher Ed Era. German industry pushes to increase working week to 40 hours. Parents Push Teens to Start Investing Earlier Than They Did. MrBeast's ‘God King’ problem. “There was no existing blueprint for becoming a toddler techno DJ.”

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!