Money Stuff

Unicorns Take Different Paths to Being Public

Also Alexa, PredictIt and Uber.

IPOs, etc.

Dropbox Inc. raised money in private markets at a $10 billion valuation in 2014. This year, it filed for an initial public offering at a valuation range of about $6.3 billion to $7.1 billion. Then demand was better than expected and it ended up going public at a valuation of about $8.2 billion. Then demand was even better and, by the close of trading on Friday, the first day that it was public, Dropbox had a public-market valuation of about $11.2 billion. It was up again -- to $11.9 billion -- yesterday.

There is a sense in which that is an awkward series of values. For Dropbox, sure: Growing from $10 billion in 2014 to $11.9 billion in 2018 is, you know, okay, but if you're locking up your money in a private venture-capital unicorn investment you probably want to make more than 20 percent over four years (about 4.4 percent annually). The S&P 500 was up more than twice as much in that period.

But it's also a little awkward for Dropbox's bankers. Private markets valued Dropbox at $10 billion. Public markets valued Dropbox at $11 billion or so. The markets were fine; the markets were on the same page. The bankers were the outliers: In valuing and marketing and pricing Dropbox, in soliciting feedback from the market and presenting it to the company, in setting the initial pricing range and the final IPO price, they got to a valuation that was well below the valuation that Dropbox's private investors had in mind, and also well below the valuation that public markets quickly came to.

I am not normally one to get too worked up about IPO underpricing. The notion that companies "leave money on the table" when their stock price shoots up after an IPO, while not untrue, is usually overblown, and most companies really want that pop. But it is more salient and annoying when the IPO pop crosses the last private valuation: Private technology unicorns prefer not to do "down rounds" where they raise money below their previous valuations; a down-round IPO is a bit of a black eye. A down-round IPO that turns out to have been unnecessary -- where the public markets quickly settle on a valuation above the last private round -- is particularly frustrating.

And Dropbox is not the only hot technology unicorn whose public-markets debut is in the news this week. There is also Spotify Technology AB, which plans to go public next week via a direct listing. Spotify's plan involves cutting back on the banks' price-setting role; instead of hiring banks to build a book of demand and set an IPO price based on the orders they receive, Spotify will just one day start letting investors buy, and letting its current shareholders sell, shares in regular transactions on the stock exchange. (Perhaps those transactions will be informed by recent private-market transactions.) The price might move around a lot, but it will go from being set by private investors to being set by public investors, without going through the banks as is traditional in IPOs. If the private and public investors are on the same page and the banks are not, then that might be an attractive outcome, for Spotify and for unicorns who come after.

The standard path for a successful private technology company used to be, you know, you raise some money at a low valuation, and then you raise some more money at a somewhat higher valuation, and then you go public at a much higher valuation and make your venture capitalists rich, and then you trade at an even higher valuation and make your IPO investors rich, and then your valuation keeps going up and your public investors also get rich. Everyone at every stage is rewarded, and in proportion to the risk they took: The early-stage VCs get very rich, the IPO investors get a nice quick pop, and the post-IPO public investors do just fine. Obviously most companies did not work out that way -- thus, the risk -- but it was the model of success.

The model of success has changed. Now the goal is not to do an IPO but to be a unicorn; now you can go around being valued at $10 billion (Dropbox) or $18 billion (Spotify) or $60+ billion (Uber), and raise billions of dollars at those valuations, without going public. But this isn't just a scaling up of the old system: It's not like the old $100 million private companies would go public at $1 billion, while the new $10 billion private companies go public at $100 billion. It's a compression of the old system: The new $10 billion private companies go public at $11 billion, or $8 billion. The late-stage private investors now are doing the job that the post-IPO public investors used to do, and are being compensated accordingly (say, 4.4 percent a year). The IPO investors -- the ones who buy in the offering -- are doing ... well, it is harder to see what job they are doing, or why they should get a 30 percent pop for doing it. Spotify may show other companies how to cut them out of the loop. Dropbox may show other companies why they should.

Speaking of cutting people out of the compensation loop, my Bloomberg Gadfly colleague Shira Ovide points out that Spotify is paying a whole ton of money to bankers for doing its non-IPO:

Spotify Technology SA disclosed on Monday that it expects costs of 35 million to 40 million euros ($44 million to $50 million) related to its coming debut as a publicly listed company. 

That's more than Dropbox paid its bankers for actually doing an IPO, and Dropbox raised $631 million.

And in contrast to a typical IPO, Spotify is getting zero dollars from its stock listing. That means whatever the company's costs for its stock market listing, it isn't functional spending. Yes, the company will benefit from stock transactions that establish a true market value for its shares, plus employees and existing investors have a chance to cash out. But by design, Spotify itself isn't selling shares in its stock market debut. That makes Spotify's costs look large compared with the value it is getting. 

It's even worse than that. When a company does a regular IPO in which it sells stock and its big shareholders also sell stock, the company doesn't pay all of the bankers' fees. The bankers don't charge "fees" at all; they get an "underwriting discount" -- notionally, they buy the stock from the company at a lower price than they sell it to the public. Everyone who sells shares pays that "discount," meaning that in an IPO where half the shares are sold by existing shareholders, those shareholders normally pay about half of the bankers' fees. But Spotify's offering is 100 percent secondary -- all of the shares that will be sold are coming from existing shareholders -- but the company will bear 100 percent of the cost.

Why is the cost so high? The obvious answer is that this offering is harder than an IPO. Spotify's banks aren't doing some of the work that IPO banks would do -- setting up roadshow visits, building a book of demand, stabilizing the deal -- but they're doing a lot of it, and whatever they are doing is brand new. As in an IPO, the banks play a role in guiding investors to a price and convincing them to buy shares, but unlike in an IPO, the banks can't rely on the investors' general knowledge of how IPOs work and how they can buy shares. Everything has to be explained from scratch to every investor. (And every salesperson.) Putting in a listing application for the stock -- normally a quick administrative task in an IPO -- required a rule change by the New York Stock Exchange. Banks are mostly paid for advising clients on doing deals; most IPOs are like most other IPOs, but every aspect of this deal requires novel and thoughtful advice.

But, you know: once. If this goes well, then it will set a template for other non-IPO IPOs; other unicorns will be able to do direct listings without reinventing the wheel. Presumably their costs will go down; Spotify will pay for a lot of work that can then be re-used for free by others. It should charge them royalties.

Artificial intelligence in finance.

There is an expression on trading floors, "3 Go 1 Go," which is the sequence of keys that you use in the Bloomberg terminal to forward a message. It is used as a dismissive term for a certain type of stock or bond salesperson, who is imagined to just sit at his computer getting messages from his traders and research analysts and mindlessly forwarding them to his customers. When I worked at Dealbreaker, a reader once sent us a photo of a fancy car with a "3 GO 1 GO" license plate. The joke was, it can be pretty easy to get pretty rich doing pretty simple information intermediation. If your trader has bonds and your customer wants bonds and you are the only point of contact between them, you can just forward messages back and forth between them and grow rich in the process.

I thought of that when I read this:

“Alexa, ask JPMorgan what the price target for Apple is.”

It’s a request that JPMorgan Chase & Co. institutional clients can now get quickly answered through Inc.’s ubiquitous voice-activated assistant. The bank and the e-commerce giant have partnered to provide JPMorgan’s Wall Street users with another way to access its research. Alexa is able to send analysts’ reports and related queries, and the bank is testing other features, like providing prices on bonds or swaps, according to David Hudson, global head of markets execution for the New York-based bank.

If your salespeople are just forwarding pricing runs and research reports to clients, then they can be replaced by a computer. If your salespeople are just in the business of pushing a button, they can be replaced by a button. 

Of course they're not. The positive spin here is not that salespeople will be replaced by virtual assistants, but that those virtual assistants will take on some of the salespeople's dumber and more administrative tasks, letting the salespeople focus on more important things, like deep meaningful conversations with clients or golf games or targeting exercises. Still there is some value to the dumb stuff too. It gives you a client contact, an excuse to talk to the client, a reason for the client to think fondly of you. If a client emails you to ask for your firm's Apple price target, you can also send her the latest research and suggest a derivatives trade that will pay off handsomely if Apple reaches that target. Alternatively, you can just quickly reply with the price target, and the client will be so grateful to you for not constantly pestering her with dumb ideas for derivatives trades that she will move more business to your bank. As a human, you can make a judgment about which approach will work better.

That might be harder for Alexa. Surely the virtual assistant can be programmed to pitch and up-sell all the time ("Alexa, ask JPMorgan what the price target for Apple is." "Before I do that, can I interest you in some nice Microsoft structured notes from JPMorgan?"), but I suspect that will be even more annoying -- and easier to ignore -- from a robot than it is from a human. And if the robot just does the job quietly and competently, it won't exactly accrue a ton of goodwill for JPMorgan. No one feels gratitude to a robot; no one feels guilty constantly asking the robot for information without ever sending it a trade. You can see why banks would want to commoditize parts of their client relationships: It's cheaper, for one thing, and also it's usually better; I'd rather get my research reports from a robot than by calling up a pushy human who wants something from me. But the banks do lose something by commoditizing relationships: Relationship businesses tend to be more lucrative than commodity ones.

In better news for the humans:

The average banker bonus in New York City was $184,220 last year, the biggest annual haul for Wall Street employees since before the financial crisis.

Those bonuses, which totaled $31.7 billion, up 17% from 2016, tracks with a broader rebound for bank stocks last year, boosted by the prospects of rising interest rates, faster growth and deregulation.

Limits to arbitrage.

Here is a fun article about a guy who left Jane Street, the quantitative trading firm that boasts that it trades more than $6.5 billion worth of exchange-traded funds every day, and went into business trading products where his position limit was $850. The products were political futures on the PredictIt market, and the limit had its perks:

While it limited how much he could get behind a bet he felt strongly about, it also created more opportunities for mispriced markets. If traders weren’t limited to $850, big bankrolls could push mispriced markets into more accurate pricing. The limits meant there would be more mispriced markets than there ought to be. 

It also affected his tactics:

Gill also discovered that once he was maxed out on a position, which he almost always was, then he had no incentive to keep his trades secret from the market. In fact, he had every incentive to shout it from the rooftops, to get others to agree with him and invest in the same position with him, and help drive the price up.

You can see this in real financial markets too. Short-selling, for instance, is hard: It is risky, and stock borrow can be expensive and hard to come by and can be called away quickly. You might think a company is a fraud and worth zero; in theory, then, you should keep selling its stock until you drive the price all the way to zero. In practice no one does this, or could do it; you can't practically just short a stock until it is priced efficiently. So you often do the next-best thing, which is to pitch your short thesis to anyone who will listen. If you're trading in a more efficient market you don't need to spend as much of your time explaining to people that you're right; the market should do that for you.


I feel like I am constantly reading stories in which Uber Technologies Inc. is abandoning its efforts to compete in some country or region, selling its business to the dominant competitor in that region in exchange for a stake in that competitor. Usually SoftBank Group Corp. is an investor in both Uber and the competitor. Here are China, and Russia, and maybe India, and now Southeast Asia, where Uber is selling to Grab:

The agreement -- - which includes all of Uber’s operations in Southeast Asia as well as Uber Eats in the region -- gives Uber a stake of between 25 percent and 30 percent in the new combined business, the people said, asking not to be identified ahead of an official announcement. The deal, which Bloomberg outlined earlier this month, marks Uber’s operational exit from yet another major market and hands a victory to Grab as it battles local competitor Go-Jek.

SoftBank Group Corp., a major backer of Grab’s and Uber’s as well as China’s Didi Chuxing, has pushed consolidation to improve the profitability of a global ride-hailing business that bleeds billions of dollars a year. 

I suppose the end game is for Uber to be a holding company for SoftBank's collection of car-hailing companies? It is sort of hard to know what the right antitrust analysis is here. In general if you have a bunch of companies that compete in the same industry and that are all influenced by a single common owner who pushes them to reduce competition and increase profits, then that seems, you know, straightforwardly bad. On the other hand you can be sympathetic to SoftBank: It is not pushing competitive businesses to raise prices to seek monopoly profits. It's pushing money-losing businesses to lose a bit less money. You can see why the owners of car-hailing companies might think there is such a thing as too much competition.

Elsewhere in common ownership, here is "Something in Common: Competitive Dissimilarity and Performance of Rivals With Common Shareholders" by Brian Connelly, Kang Lee, Laszlo Tihanyi, Trevis Certo and Jonathan Johnson:

We examine the competitive and performance implications of common institutional ownership, which occurs when an institutional investor owns a sizeable number of shares in two publicly traded firms. We argue that rival firms with common ownership structures will engage in dissimilar competitive action repertoires to avoid direct competition with each other. Competing aggressively, but with dissimilar action repertoires, allows rivals to maintain high levels of performance. Competing with dissimilar action repertoires also helps ensure that performance disparity between the two firms remains low. In other words, competitive aggressiveness with dissimilar action repertoires yields an optimal competitive solution for rival firms and their common owners.

And elsewhere in Uber: "Arizona Governor Suspends Uber’s Self-Driving Cars From Roads."

Things happen.

Deutsche Bank Seeking Replacement for CEO, London Times Says. Ackman Vows to End Three Years of Underperformance. Avon Gives Activist Group Board Seat, Avoiding Proxy Fight. Brookfield Strikes Deal to Buy Rest of Mall Owner GGP. GSK Gets Full Control of Sensodyne Maker for $13 Billion. How a Tiny Latvian Bank Became a Haven for the World’s Dirty Money. White House Lawyers Looking Into Loans to Kushner's Business. Homeowners Ditch Refinancings as Mortgage Rates Rise. Facebook is under federal investigation. Flying Whale’ Blimp That Never Lands Joins Global Airship Race. 'Billions' creators discuss their charmingly devious characters who have everyone on Wall Street saying, 'You know that was based on me, right?' UVA Knocks Off Defending Champ UMBC to Reach National Cybersecurity Competition. Fired Vancouver waiter: I'm not rude, just French

If you'd like to get Money Stuff in handy email form, right in your inbox, please subscribe at this link. Thanks! 

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

    To contact the author of this story:
    Matt Levine at

    To contact the editor responsible for this story:
    James Greiff at

    Before it's here, it's on the Bloomberg Terminal.