Money Stuff: Bankers’ Hours Are Still Pretty Bad

I guess the old standard for investment banking analyst hours was “it’s about 100 hours a week, unless you need to work more,” and the new standard is “it’s 80 hours a week, unless you need to work more”:

JPMorgan Chase & Co. will limit junior banker hours to 80 per week in most cases, according to a person with knowledge of the matter. Exceptions may include extra work to complete live deals, the person said.

Doesn’t that still sound bad? Like an 80-hour week is pretty long, and when you phrase it like that — “80 hours per week unless there’s a live deal” — that suggests that it’s a floor, not a cap. If there is nothing particularly urgent for you to do, it’s fine for you to come in at 9 a.m. and leave at midnight Monday through Friday and work just one afternoon per weekend, but obviously if you’re on a deal you’ll need to do more. Eighty hours a week is a lot to work when there isn’t anything urgent.

Meanwhile Bank of America Corp. has had a problem where (1) it has an official policy of limiting junior bankers’ hours but (2) the midlevel bankers who actually control those junior bankers’ lives apparently did not believe in the policy. The result was that the midlevel bankers would tell the junior bankers to report working fewer hours, so that human resources would not stop them from working more, and the junior bankers would do that. Since this problem became public, Bank of America has sent out memos saying “no actually we really mean the policy,” and now it is also changing how the hours are entered:

Some of its junior bankers described to Bloomberg in June how they were shaving down their reported worktime to avoid breaching 100-hour weekly limits. They worried that crossing that line would trigger questions from human resources and draw the ire of managers.

This month, the Charlotte, North Carolina-based firm is rolling out a new system for monitoring trainees’ work, with hours reported on a daily basis, rather than weekly, according to a person with knowledge of the matter. Part of the idea is to spot who’s busiest and who else has capacity, spreading assignments.

If you’re asked to report your hours once a week, and you worked 120 hours, and 101 is too many, you will report 99. But if you report once a day it’s a little harder to fudge.

These stories are all a little strange, because there is another popular genre of stories about junior bankers, the one that is like “banks and vendors are rolling out artificial-intelligence-driven systems that will spot deal ideas, build financial models, and prepare pitchbooks with almost no human oversight.” Shouldn’t that make the analysts’ lives easier? Like you can roughly think of the work of a junior investment banker as:

1. Building financial models for mergers and leveraged buyouts, using public data, some knowledge of accounting, some financial and industry understanding, some sense of markets, and some pretty good skill with Excel shortcuts.
2. Formatting pitchbooks and putting all the logos in the right places.
3. Running processes and setting up data rooms and meetings for live deals.
4. Sometimes, as a treat, meeting clients and learning how to interact with them and sell to them.
5. Absorbing all of this, so that, after two years, the analyst knows a lot about the companies she covers and has a deep intuitive grasp of what items drive changes in the financial model, so that she can give clients informed off-the-cuff financial analyses without an Excel model, or suggest some acquisition ideas on the spot without a pitchbook, or give a client informed tactical advice on how to run a merger auction.

The traditional reason that investment bankers work long hours is that the first three items — financial modeling, making pitchbooks, and doing grunt work on live deals — take a long time, and somebody has to do them, and if it takes until 4 a.m. then it takes until 4 a.m.

That was unpleasant, but it came with the nice side effect of accomplishing the last item: If you work 100 hours a week for two years building LBO models, you will internalize things about LBO economics that you can’t get by just reading about them. If you work 100 hours a week for two years writing pitchbooks, you will develop a deep sense of what sorts of deals get pitched and what pitches work. The knowledge and intuition necessary to do the job will be pumped into you at great speed. “Drinking through a firehose” is the cliche. 

And then you eventually graduate to being a senior banker, or more realistically a private equity investor, and you are useful. You know stuff, because you learned it, fast.

And now we’re pretty close to a world where an AI can do most of those first three items, build the models and write the pitchbooks and put the stuff in the data room and so on. You could imagine that having the effect of making junior bankers’ lives easier in the short term, but undermining their development in the long term: They won’t learn, at a deep level, what drives returns on an LBO, because the AI does all the math for them. The apprenticeship model will break down.

But so far you have to imagine that because instead the junior bankers keep working 100 hours a week. Shouldn’t the AI at least be able to help with that?

Rupert Murdoch’s family trust owns about 14% of the stock of News Corp., but News Corp. has dual-class stock, and Murdoch owns about 41% of the voting stock. There is a similar dual-class structure at Fox Corp., where Murdoch controls about 43% of the vote. This is controversial — we talked the other day about an activist News Corp. shareholder trying to get rid of the dual-class stock — but not uncommon. Particularly in the media business, there’s a sense that the founding family ought to have control over the company, insulated from the pressures of the market. The “dual-class capital structure promotes stability and has facilitated the successful implementation of News Corp’s transformational strategy,” says News Corp.

When Murdoch dies, his six children will each inherit a portion of the economic ownership of that stock. Right now four of those children stand to inherit equal voting rights over the trust, and they seem to be at odds with each other about the company’s business strategy and journalistic standards. This arguably undermines the purpose of the super-voting stock: You might want one person to have control of the company, insulated from conflicting pressures, to “promote stability.” Murdoch might want the economic ownership (of his 14% ownership stake) to be divided among his children, but he might want the voting control (of his 41% voting stake) to be concentrated in one of them.

And so the solution is dual-class stock on top of dual-class stock: Give one Murdoch child (Lachlan) super-voting control over the trust that has super-voting control over News Corp. and Fox Corp.:

Rupert Murdoch, 93 years old, is seeking to amend his trust—which holds big stakes in Fox News owner Fox Corp. and Wall Street Journal parent News Corp—to ensure that when he dies Lachlan will control the family holdings spanning cable news, sports media, streaming, newspapers, book publishing and real estate. James Murdoch is resisting the change, and these days he has Elisabeth and Prudence on his side. 

A trial to settle the dispute is set to begin Monday in a probate court in Nevada, where the trust is based. It will be closed to the public unless pending legal challenges to open up the proceedings are successful.

The Murdoch trust exercises considerable influence over Fox and News Corp, with roughly 40% voting stakes in each company. Rupert now controls it, but when he dies, each of his four oldest children will have an equal voting share, people familiar with the matter said.

For Rupert Murdoch, putting more power in Lachlan’s hands is meant to ensure stability at the businesses and avoid a confusing ownership structure in coming years, associates of the mogul said. If he prevails, Lachlan would have the same sweeping authority his father has enjoyed for years, and other shareholders would largely have to accept whatever his plans are for the companies. …

Under the proposed changes, Lachlan would eventually gain control when Rupert’s votes go away. James, Elisabeth and Prudence would still remain financial beneficiaries, alongside two younger siblings from Rupert Murdoch’s marriage to Wendi Deng, Grace and Chloe. 

I don’t know the economic division, but if you assume that Lachlan Murdoch gets one-sixth of the economics of the trust, then he’d have economic ownership of about 2% of News Corp., while controlling 41% of the vote. 

I guess you could keep doing this? He could divide his ownership stake among his children, while giving one of them voting control, super-voting control over his super-voting control over the trust’s super-voting control of News and Fox. Possibly good for stability, but weird for alignment of incentives. 

What are the costs of capital of these two things?

OpenAI is in talks to raise $6.5 billion from investors at a valuation of $150 billion, according to people familiar with the situation.

The new valuation, a figure that doesn’t include the money being raised, is significantly higher than the $86 billion valuation from the company’s tender offer earlier this year, and cements its place as one of the most valuable startups in the world.

At the same time, OpenAI is also in talks to raise $5 billion in debt from banks in the form of a revolving credit facility, said one of the people, all of whom asked not to be identified discussing private information.

That is:

1. If you are a venture capitalist (Thrive Capital is leading the round) or big tech company (Microsoft Corp., Apple Inc. and Nvidia Corp. are all in talks to invest), what sort of return do you expect to get on this investment? As just a wild stab at it I will guess something like “you expect OpenAI to be a $1 trillion public company within 10 years,” in which case you’d get back about 7x your investment, for an annualized return of about 21%. I feel like that would be a slightly disappointing outcome? We talked recently about OpenAI’s fretting that limiting investors to a 100x return on their money is too restrictive; a trillion dollars in 10 years is, by those standards, a modest goal. Anyway if it works out this way then the investors’ $6.5 billion investment will pay out about $43 billion in 10 years.
2. If you are a bank lending $5 billion to OpenAI, what interest rate would you charge? Five-year SOFR swaps are around 3.15% today. I guess you’d charge more than that. Not … not much more than that? Like OpenAI probably doesn’t have tons of factories that you can seize and sell, but a $5 billion senior claim on a company with a $150 billion equity value and, you know, all this hype does seem pretty safe. Plus, let’s be clear, if you are a bank lending money to OpenAI right now, you are not just doing it because it’s a good loan. You’re doing it because you hope that OpenAI will eventually do an initial public offering, [1] and you want to get the glory and fees of leading that underwriting. Lending them money now is a way to position yourself for that, and the economics of the loan don’t matter all that much.

Obviously if OpenAI fails then the cost of all that venture investment will be zero (it doesn’t need to be paid back), while the cost of the loan will be high (it does), but OpenAI strikes me as pretty confident these days.

One model that you could have of OpenAI is that it started as a nonprofit organization trying to build artificial intelligence for the benefit of humanity, and quickly ran into the problem that building AI is really expensive. “It became increasingly clear that donations alone would not scale with the cost of computational power and talent required to push core research forward,” says OpenAI.

So it started raising money [2] from investors in a way that looks a lot — not entirely, but a lot — like traditional venture capital financing of a hot tech startup. There is weird governance, with a nonprofit board that is not technically beholden to investors, and the investors’ potential returns are capped (with a pretty high cap), but basically the proposition to funders changed from “give us money and, if we succeed, you can feel good about doing something good for society” to “give us money and, if we succeed, we will give you back just so much more money.” [3]

And that shift probably did make OpenAI the juggernaut it is today, to the point that … I’m not even sure that pitch is necessary anymore? At this point OpenAI is so hot that “give us money and we will spend five minutes meeting you to accept your money” is a plausible pitch. “Lend us money and we’ll probably pay you back, and you can say you’re one of our lenders” is a thrilling pitch. 

Loosely speaking, an index fund is a passive investment fund that doesn’t try to pick the good stocks but just buys all of them. More accurately speaking, though, “buy all the stocks” is not all that practical a rule. All of them? What about stocks with tiny floats, or stocks that only trade three shares a day? Many US index funds are based on the S&P 500, an index of roughly 500 large-cap stocks, which represents much of the market (measured by dollar value) but omits thousands of companies. 

You can get closer. [4]  Perhaps the closest you can get in the US is the Vanguard Total Stock Market exchange-traded fund and index fund, each of which “seeks to track the performance of the CRSP US Total Market Index,” which in turn “represents approximately 100% of investable companies in the U.S. equity market.” Approximately 100%.

CRSP — the Center for Research in Security Prices, which provides that index — points out that it is “comprised of more than 3,500 constituents across mega, mid, small and micro capitalizations,” ranging from $2 million to $3.3 trillion. But it doesn’t actually capture every stock, every day. Here is CRSP’s methodology guide, describing how its index is made. There are various intentional omissions: “Pink sheet” stocks that are publicly traded but not listed on a stock exchange, for instance, are not eligible for the CRSP index; nor are stocks that don’t have sufficient trading volume or float; nor are most investment companies and special purpose acquisition companies. [5]

But not even every eligible company is in the index the moment it becomes eligible. “CRSP Market Indexes are reviewed on a quarterly basis,” so a company that is too teeny or thinly traded to be in the index, but that becomes bigger and more actively traded one January, will generally be added to the index that March. In February, the index is not quite 100% of the investable market.

And of course private companies aren’t in the index: They don’t trade publicly, so they are not “investable” for Vanguard. But sometimes private companies go public, in an initial public offering, and then they get added to the index. For some more famous indexes, there is a long delay between going public and being added to the index, but CRSP is often quicker: Companies that IPO get added to the CRSP index next quarter, except that relatively big IPOs are “fast tracked” and added to the index “after the close of the fifth day of trading” after the IPO. (“Relatively big,” for CRSP, means “the company’s total market capitalization is at least as large as the breakpoint of the CRSP US Small Cap Index,” which is a market cap of a bit less than $2 billion. Below that breakpoint are roughly 1,800 companies in the micro cap index. Most IPOs don’t qualify for fast tracking, [6]  because they are teeny, but pretty much all of the IPOs that you’ve heard of do. [7] )

So very loosely speaking the total stock market index fund buys “all of the investable stocks,” but when a stock first becomes investable, the fund doesn’t immediately buy it: It either waits until the next quarter to buy it, or — for a fast-track IPO — it waits five days to buy it.

If you are a private company thinking about an initial public offering, you will have to find investors to buy your stock in the IPO. Most of those investors will be institutional investors who listen to your story, decide they like it and put in an order. Others will be retail investors whose financial advisers get them into the deal. But if you are big enough for the CRSP fast track, then some of your IPO investors will be CRSP index investors (basically Vanguard funds). Oh, not really: You won’t be in the index for five days after the IPO, so the index funds won’t actually put orders into the IPO. But their demand for shares is so predictable that somebody will buy in the IPO to flip to them, and then the stock will trade up a bit after the IPO, because of that bump of index demand. Here’s a fun paper by Marco Sammon and Chris Murray of Harvard:

We document the effects of mechanical buying by CRSP-index-tracking funds on post-IPO returns and IPO deal structure. Leveraging a difference-indifferences style design built on a 2017 CRSP rule change, we find that expected index fund demand leads fast track IPOs to outperform non-fast track IPOs by 15 percentage points shortly after the IPO, although this outperformance largely reverts within six months. Further, fast track IPOs are priced higher and are more likely to be upsized, raising 7.7% more capital than similar non fast track IPOs, evidence that expected passive buying has real implications for firms raising capital in public markets. 

Basically before 2017 only mid-cap or larger IPOs got fast-tracked; after 2017 small-cap IPOs did too, so you can isolate the effect of the fast-tracking. [8] Those effects are large:

Our baseline estimates suggest that fast track IPOs outperform their peers by 15 percentage points from the IPO itself to the close when CRSP-tracking funds are expected to buy. This is an economically large effect, on the same order of magnitude as the average return from the IPO itself to T+4 over our sample of 25%. ...

Our estimate of average expected buying by CRSP-tracking index funds is 7% of the recently IPOed firm’s shares outstanding. This is an economically huge demand shock, larger than the net buying by index funds after e.g., Russell or S&P Index family additions. 

One odd point here is that the authors find both effects: The IPO raises more money at higher prices if everyone knows that index investors will show up in five days, and the stock trades up over those five days. The IPO process reacts to the looming presence of index funds, but it underreacts.

I enjoyed (and blurbed) The Trading Game, Gary Stevenson’s rather dark memoir of his time as a foreign exchange trader at Citigroup Inc., though I confess that I was a bit puzzled by his frequent claims that he was the most profitable trader at Citi. I enjoy some bluster, though, and didn’t think too much about it. Here, however, is a Financial Times investigation of that claim, sourced to “eight former employees of Citi who worked with Stevenson at various points in his career,” all of whom were skeptical:

More than one of his former colleagues on the trading desk alleged that Stevenson had “delusions of grandeur”, while several said they doubted his record would have put him in the top 10 in Citi’s FX division at any point.

As a former investment banker whose record probably never put me in the top 10 by revenue on my dozen-person desk, I am fine with this, but the investigation is very thorough.

UniCredit’s Orcel Says Commerzbank Takeover Is an Option. China Detains Investment Bankers, Takes Passports in Corruption Sweep. Goldman Sachs to earn $92mn from sale of Pringles owner. Private-asset model portfolios. The AI bill driving a wedge through Silicon Valley. Brussels explores Draghi option of extending up to €350bn in EU debt. Don’t Take Advice From a Habsburg. Ragebait. 

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!