Vision A pretty normal thing to do, if you are a venture capital fund, is to invest in a private tech startup, watch it grow and go public, cash out the investment by selling your shares to the public (either in the initial public offering, or a few months later), and return the cash to your investors. That is basically what SoftBank Group Corp.'s Vision Fund is doing with its Uber Technologies Inc. investment, except that it's SoftBank, so it's doing it in a slightly weird way: SoftBank's Vision Fund plans to borrow $4bn against its stakes in Uber and two other Silicon Valley groups, as the investment arm of the Japanese conglomerate looks to hand back cash to investors in its $100bn fund. The fund is in discussions with banks, including Goldman Sachs, to arrange a loan secured against its holdings in Uber, which listed this month, and Guardant Health as well as its stake in soon-to-be-floated Slack, according to people directly involved in the deal. If the value of those stakes falls beyond a certain threshold, the fund will be obliged to stump up more cash, the people said. … By borrowing against the value of publicly traded shares, SoftBank can return capital to its investors in the fund while bypassing "lock-up" periods that prevent shareholders from selling stakes in newly-listed businesses. Sure, why not. SoftBank owns about 13% of Uber, worth about $8.9 billion, plus about $2.1 billion of Guardant and perhaps another billion dollars' worth of Slack. If you put up $12 billion worth of stuff as collateral for a $4 billion loan, then it's a pretty safe loan. Your banks will probably get paid back. If all the stuff loses 50% of its value, a fairly catastrophic drop, there's still more than enough stuff left to pay back the loan. Also, because it seems to be a non-recourse loan,[1] if the stock loses 100% of its value then you can still keep the $4 billion. And so in most scenarios you don't have to think of it as a loan. It's just: SoftBank has sold the first $4 billion of its $12 billion of stock in these companies. It will sell the rest as it becomes available, or as market conditions dictate or whatever, for more or less than $8 billion, but for now it is cashing out $4 billion in exchange for some of its shares.[2] This makes sense: These are (or soon will be) mature-ish public companies, and it's time for the Vision Fund to recycle the capital that it put into them and use it for other investments, or return it to its investors so they can invest it in the next Vision Fund. It does seem a bit like cheating when you put it like that, though? Uber Technologies Inc. went public a few weeks ago, and early investors like SoftBank signed lock-up agreements saying that they would not "offer, pledge [or] sell" their stock, or "enter into any swap, hedging transaction, or other arrangement that transfers to another, in whole or in part, any of the economic consequences of ownership" of the stock. Economically, a non-recourse margin loan looks a bit like a hedging transaction: It is sort of an out-of-the-money put option, in the sense that if the stock falls really far then the banks bear some of the loss. (SoftBank gets to keep the $4 billion even if the stocks go to zero.) And legally, a margin loan seems to be a "pledge," in the sense that SoftBank is putting up the stock as collateral for the loan. But you can, I guess, structure around that: "A person close to the deal said that the margin loan features, whereby collateral had to be posted to the banks on a share price drop, would kick in after the final lock up period on the underlying shares expires," sure. (Until then they are, I suppose, collateralized only by the promise of future collateral.) And it seems to be a thing: A lawyer said that margin loans have been a common feature of the recent rush of tech US listings, often as a way of early investors to cash out some money ahead of the expiry of lock-up periods. "Shareholders are considering it more and more as either a way around lock-up, to put on more leverage or to maximise their proceeds," a banker said. We have talked a few times recently about shareholders looking for, and sometimes finding, ways around lock-ups. Here's one more. Like other ways around lock-ups, it seems a little bad for the company. The lock-up is a way of signaling that the company's big early shareholders are committed for the long term; when those shareholders dump or semi-dump or even just margin their stock a few weeks into the lock-up, that clouds that message a bit.[3] But shareholders seem to want porous lock-ups, and companies seem to want strict ones, and right now the power seems to be tilted toward the shareholders, due to differences in negotiating leverage or perhaps just in legal cleverness. Still there are much simpler ways around lock-ups. For instance, there is Slack Technologies Inc., which SoftBank is also plopping in its margin-loan basket. Slack's IPO just won't have a lock-up at all. Because it is not an IPO, it's a direct listing: Rather than coordinating to sell a bunch of stock at once and then no more for six months, Slack will just list on an exchange and let anyone sell whenever they want. It seems a bit more straightforward than having a lock-up and making shareholders get creative to get around it. Blockchain blockchain blockchain The way you trade financial stuff is that A agrees to buy some stuff from B—maybe over the phone, or electronically on an exchange—and then A updates its computers to show that it owns the stuff, and B updates its computers to show that it sold the stuff. Unless A and B are very special people, like the Federal Reserve or something, this is not enough to really show ownership of the stuff, though.[4] There will probably be some central record-keeper—an exchange, a clearinghouse, a depository, a central bank—who keeps track of who owns all the stuff, and A and B will have to ping the record-keeper to formally transfer the ownership. And in practice there will likely be some extra intermediaries involved too. A and B might keep their stuff with custodians, and they'll have to ping the custodians to update the ownership. You'll probably have to update five or more people's computer systems to reflect that A bought the stuff from B. The way a blockchain works is that A agrees to buy some stuff from B, and then they ping everyone to tell them to update their computers to reflect that A owns the stuff and B doesn't. "Everyone" is a somewhat technical term there; for a lot of financial-industry blockchain projects, there is some group of participants—all the major banks who trade a particular asset, say—and you have to get them all to update their records. Doing it this way has some big advantages over the traditional method. Everyone keeps their records in the same format and updates them all at the same time in the same way, and so they spend less time disagreeing and reconciling after the fact. Communication between these systems is pretty efficient; everyone is using the same system. And because everyone keeps all the records simultaneously, the system doesn't rely too much on a centralized record-keeper, which builds in some redundancy: The system isn't as vulnerable to a hack or mistake or fraud or computer failure at a single central party. On the other hand, just, like, if you have to update 20 or 30 or 200 or whatever systems each time you trade, that's going to take longer than updating five. It's just math! I am constantly reading claims, not just that blockchain is a smarter or better or more secure or more open or whatever way to keep track of financial assets, but that it is faster. As far as I can tell this claim is based mostly on the fact that "blockchain" sounds computer-y, and so seems like it should be fast. But in fact having everyone in the world agree on a trade should take longer than having just the counterparties to the trade—and yes fine a clearinghouse and a few custodians—agree on it. Also the way you agree on a trade in the traditional system is, like, you write it down in a spreadsheet; the way you agree on a trade in the blockchain system is that you do a ton of cryptography and then write it down. Anyway: A trial project using blockchain to transfer and settle securities and cash proved more costly and less speedy than the traditional way, Germany's central bank president said. The experiment, launched by the Bundesbank together with Deutsche Boerse in 2016, concluded late last year that the prototype "in principle fulfilled all basic regulatory features for financial transactions." Yet while advocates of distributed ledger technology say it has the potential to be cheaper and faster than current settlement mechanisms, Jens Weidmann said the Bundesbank project did not bear those out. "The blockchain solutions did not fare better in every way: the process took a bit longer and resulted in relatively high computational costs," Weidmann said in Frankfurt on Wednesday. "Similar experiences have been made elsewhere in the financial sector. Despite numerous tests of blockchain-based prototypes, a real breakthrough in application is missing so far." This is not even an objection really! You could imagine a lot of benefits to the blockchain, in terms of security and redundancy and also secondary efficiencies of having everyone keep copies of the same data in the same format. It's just, you know, if you think "having lots of people do a ton of computation will be faster and cheaper than having a few people do a little computation," you are kind of coming at the blockchain wrong. Payment for crypto flow Here's the basic way that "payment for order flow" works. Many electronic high-frequency trading firms follow more or less market-making strategies, where they try to buy stocks for $9.99 when people want to sell, and sell them for $10.01 when people want to buy, and balance their buys and sells to make lots of tiny profits without much risk. If you do this in the public markets, you run the risk of what is technically called "adverse selection," or less technically "getting run over": You might trade with someone who knows something you don't, or who is just in the middle of buying or selling a ton of stock, and the price might rapidly move against you. If you can prevent that—if you can screen out big institutional investors and smart hedge funds—then your risk of adverse selection goes way down. So retail brokerage firms have something valuable to offer high-frequency traders: Their customers' orders tend to be small, random and uninformed. HFTs will compete to trade with those retail brokers' orders, which carry so much less risk. They will compete by offering price improvement: They will trade with retail orders at tighter bid/ask spreads (paying more to buy and charging less to sell) than the public markets, because the risk is lower. They will also compete by paying for order flow: They will give money directly to the retail brokerage in exchange for its orders. Being a market maker in cryptocurrency seems hard! It's super volatile, so if prices move against you they will probably move against you by a lot. It's relatively unregulated, and there are constant claims about price manipulation and insider trading, as well as rumors about big "whales" dumping their currencies and moving prices. The risk of adverse selection, in an unregulated and confusing market, seems high. But if you could trade with (1) retail customers (2) who are trading cryptocurrency on their phones and (3) who only got into cryptocurrency trading fairly recently, then your risk goes way down. Someone who installed a crypto app on his phone six months ago is unlikely to be a whale or a manipulator. You should be willing to pay a lot for those customers. Anyway: Robinhood Markets Inc.'s move to add cryptocurrencies trading last year was certainly a popular one, helping the maker of the free trading app double its customer base. But it needed help to pull it off. With many of the major trading houses wary of crypto, the question remained how the fintech firm was going to buy and sell Bitcoin, Ether and other digital assets on behalf of customers. For that, it's turned to Jump Trading LLC, a little-known but profitable Chicago trading firm that's also benefited from being an early mover on cryptocurrencies, according to two people familiar with the matter who asked not to be identified discussing private matters. ... For Robinhood, the crypto offering helped its efforts to court younger investors and pushed it above $100 billion in transaction volume, which helped lead to a valuation of $5.6 billion last year. The venture capital-backed startup makes no secret about sending crypto orders to trading venues for execution, just as it does when a user buys or sells a stock on a smartphone. But unlike stock trading, in the less-regulated world of digital assets it isn't required to say where it sends orders. According to Robinhood's website, the app company receives "revenue from these trading venues in the form of volume rebates. These rebates help us cover the costs of operating our business and offer you commission-free trading." There is, as I say periodically about Robinhood's payment-for-order-flow setup, nothing wrong with this. If Robinhood sends your order to a venue that (1) gives you good execution and (2) pays Robinhood to subsidize its free trading, then you have no reason to complain. But there is a sort of meta-problem with it, which is that the setup here is premised on the idea that your order is a mistake. "You shouldn't be trading crypto on your phone" is kind of—not entirely, but kind of—the premise behind the whole operation; if you should be doing it, if you had some advantage, then no one would want your orders. The fact that they do want your orders means that you get good execution, but if they're right then that good execution might end up being bad for you. Maybe you should listen to them! Lunch valuation Well here we go again: A charity auction to have lunch with Warren Buffett has hit a record before it is even over. Bidding for the annual lunch auction reached $3,500,100 late Monday, one day after the auction opened at $25,000. The auction to meet the billionaire and Berkshire Hathaway Inc. chairman ends Friday night. All bidders have to be prequalified to submit an offer. The auction is run by eBay and auction manager Matchfire. An anonymous bidder set the previous record high of $3,456,789 in 2012, and that amount was matched in 2016. Last year's winning bid, also anonymous, was $3,300,100. Berkshire's stock price has more than doubled since 2012, so by the rules of lunch valuation, $3.5 million is a steal. It's a little weird that this is such a big, expensive, once-a-year thing. It seems safe to assume that Warren Buffett eats more than a thousand meals a year. Presumably he eats at least several of them with other people: friends, colleagues, investment bankers pitching him deals, etc. If lunch with Buffett is really a multimillion-dollar commodity, shouldn't there be, like, a little bit of a black market in it? If you are an investment banker having lunch with Buffett to pitch him a deal, or just an old friend of his having dinner to catch up, wouldn't you be tempted to bring along an associate or a friend, and sell the spot? I like to imagine there's some investment-banking VP out there who didn't get a bonus last year, and her boss was like "well but you went to that Buffett meeting, that's worth like a million bucks, there's your bonus." Or why doesn't Berkshire monetize this? Figure he eats 1,095 meals per year; at $3.5 million each that's $3.8 billion of revenue that Buffett and Berkshire are just leaving on the table. I suppose that is not how it works. Probably most of that $3.5 million price tag comes from the scarcity and publicity value of the annual auctioned lunch; if he was selling off every Tuesday breakfast then the rate would be lower. Still it seems implausible that the revenue-maximizing frequency is one meal a year, though revenue maximization is probably not the main goal. Things happen New Jersey's Pension Fund to Cut Hedge Fund Allocation in Half. Judge Denies Ratepayers' Bid for Representation in PG&E Bankruptcy. WeWork Is Said to Be in Talks for $2.75 Billion Credit Line. Uber's First Earnings Will Test How Well It Manages Expectations. Uber is now kicking low-rated passengers out of its cars. A $4 Trillion Plan Could Make or Break Dreams of U.S. Homebuyers. The Department of Energy Is Now Calling Fossil Fuels "Molecules of Freedom" and "Freedom Gas." 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! [1] "SoftBank has favoured margin loans because even though banks can seize the underlying stock if it falls heavily, they cannot go after any of the borrower's other assets." [2] Well, in exchange for a variable number of its shares: If the stakes are up when the loan comes due, the Vision Fund can pay off the loan by selling fewer than one-third of its holdings; if they're down, it will need to sell more than one-third. (Or it can *not* sell the shares and come up with the cash to pay the loan in other ways, but why do that?) [3] Also there is the possibility that the margin loan counterparties, who have stock price risk (because if the stock drops really far they don't get paid back), will hedge that risk by selling stock short. [4] Try it! If I agree to sell you the Brooklyn Bridge, and you agree to buy it, and I make a note in a spreadsheet showing that I sold it, and you make a note that you own it, that will not suffice. | 
Post a Comment