The WeWork Arc

Plus! Amazon’s Third-Party Data; The Lucas Critique in Finance; Matching the Huawei Strategy; Froth; Demand and Logistics

I decided to make this issue of The Diff free. WeWork has been a fun story for years, and it’s finally gotten a book-length treatment. The story is still widely misunderstood, but now it’s been told from start to finish.

In this issue:

  • The WeWork Arc

  • Amazon’s Third-Party Data

  • The Lucas Critique in Finance

  • Matching the Huawei Strategy

  • Froth

  • Demand and Logistics

The WeWork Arc

WeWork is an ideal company for a business book. Per my general theory of business books, the ideal recipe for a satisfying narrative about business is:

  1. A classic Greek tragedy, where the hero is undone by his own hubris, and

  2. Lots of people with free time to talk to an author, who have a vested interest in telling their side of the story.

WeWork has both. The company’s financial history sounds like an extended roulette session: every year, the company doubled in size, until 2019, when it shrank to almost zero. And the story is tied to the ambition of a single founder, Adam Neumann, whose sales ability and indifference to risk propelled the company to a $47bn valuation and then led to its near-collapse.

WeWork got a lot of media coverage, slowly on the way up and then much more frequently on the way down, and now the story has been told in the just-published Billion Dollar Loser.

One thing the book’s narrative makes clear is that WeWork was not just a creation of the venture capital market of the late 2010s. It was also a creation of the labor and real estate markets of the early 2010s. WeWork’s founders, Adam Neumann and Miguel McKelvey, started a predecessor company called Green Desk in early 2008, leasing office space in a building in Brooklyn and subleasing smaller units. (In a memorable exchange, Neumann pitched this idea as a way for his landlord to get some use out of vacant space. The landlord said “You know nothing about real estate,” and Neumann replied “Your building is empty. What do you know about real estate.”)

If Green Desk was a pure real estate arbitrage, it had the worst possible timing. Green Desk committed to paying for commercial real estate at close to the market peak, and would have to resell it into the softer recessionary market. As it turns out, it was an “arbitrage” the same way a restaurant arbitrages the price gap between raw ingredients and a fully-prepared dinner: Green Desk’s customers were thrilled to buy real estate in the form they actually wanted, rather than signing longer leases for bigger spaces than they could afford. The early operation was scrappy—a negotiation with Gothamist, an early tenant, was resolved when Green Desk agreed to buy a table for the office conference room if Gothamist found chairs. And the company didn’t have to pay early employees much at all. By 2009, Green Desk was successful enough that the landlord bought the founders out and decided to expand the concept himself.

The founders turned around and launched a competitor, WeWork.[1] WeWork had to pitch landlords and tenants. Tenants were easy—recessions make people leery of long-term commitments, and layoffs-with-severance are a way to force-feed startups their seed funding. Landlords were harder to pitch, because their question was: why can’t we do WeWork ourselves? Miguel McKelvey said after the fact WeWork didn’t have a particularly good answer to this question, but apparently Adam Neumann did, because the company was able to sign leases—and was able to score an early angel investment, at a $45m valuation, before they’d done gotten first property.

Once WeWork started growing, the pace didn’t slow down. The company got good at managing both capital needs and tenant expectations; they found that they could extract rent-free concessionary periods from landlords based on how long it would take them to set up a new location, and then finish the setup process faster. While most commercial tenants don’t like to move into an unfinished building, WeWork’s customers were ready as soon as there were chairs and WiFi.[2]

In the depressed real estate market of the post-crisis period, WeWork could at least find landlords who were willing to hear their pitch, and could also find employees who were willing to work for below-market salaries. WeWork promised stock options for years before offering them to employees, for example, which kept their costs low. Meanwhile, WeWork had started raising funds from venture capitalists. WeWork was consistently able to push for a higher valuation than comparable real estate companies by a) pitching itself as something closer to a tech play (at first an in-person social network, then as the hype cycle changed, a space-as-a-service company), and b) by growing fast.

WeWork’s model started to break because there’s a tension between its two competitive advantages. Underpaying employees and overhyping office space is good for margins, and overselling investors is a good way to fund growth, but after a while, the only way to grow is to overpay in order to hire employees and sign leases.[3] And WeWork’s valuation was, for a while, sustained by its topline growth; it had to sacrifice either access to capital or margins, and it chose margins. If WeWork’s competitive advantage a CEO who excelled at sales, this actually made some business sense: the person who can pitch investors on a million-dollar investment can pitch different investors on a billion-dollar investment, but the same person can’t negotiate every lease when the company is opening two new locations every day.

The simplest explanation of WeWork is that the company ran out of people to fool: they used money from small investors to raise money from bigger ones, until they finally tapped out Softbank, the world’s biggest investor in private tech(-ish) companies. An IPO provides liquidity, but it’s also the first time short-sellers can participate in price discovery, which means it’s a tough experience for controversial companies. (As one member of WeWork’s finance team put it: “The nature of private markets is that if nine smart investors pass, it only takes one relatively dumber investor, and suddenly we’re valued at $16bn.”)

But that greater-fool explanation doesn’t quite hold water. WeWork’s initial capital came from the founders' successful real estate dealings, and the first outside money came from another real estate investor. And they raised venture money from some sophisticated counterparties—even Softbank, which offered them a ridiculous valuation, was careful to structure the deal so it protected them from some downside risk.

And, strangest of all, the company didn’t go to zero. They fired Neumann (proximate cause: adverse PR from smoking marijuana on the company plane while traveling internationally; actual cause: inability to complete an IPO).[4] They laid off thousands of employees and sold some assets. And WeWork’s new CEO says it will be profitable in 2021 ($, FT), the most impressive transition from disaster to smooth landing since Apollo 13. WeWork’s skill at fundraising turned out to be great for the early investors, and only a disaster for the last and biggest. Its talent strategy changed from underpaying at a small scale to overpaying at a large scale, so the company’s economics when they filed their prospectus overstated how unprofitable the undelrying business was. And while a coworking company is a terrible business in the middle of a pandemic, it’s a great business for a post-pandemic world where, as in the early 2010s, most companies are cautious about committing to space and willing to accept unconventional office arrangements. If WeWork hadn’t already existed, the rise of partially-remote work and the collapse of commercial real estate prices in big cities would have necessitated its invention.

The personality traits necessary for quickly building a property business have a negative correlation with the traits necessary to build one that doesn’t collapse. Neumann comes across as absolutely magnetic in the book; my big regret was that there wasn’t a camera crew following him around from day one. Which raises an interesting point: maybe the sorts of people who can build a big, rickety real estate empire really ought to be in the high-margin, recurring revenue-driven business of reality TV. We already have one case study of this, and Neumann is only 41. He has time to reinvent himself again.

[1] In what’s now an amusing sidenote, the Green Desk sale included a noncompete: the founders couldn’t operate a coworking space in Brooklyn, giving them no choice but to try the vaster market of Manhattan. Just like France reached its imperial peak under a Corsican emperor and Russia under a Georgian, Manhattan’s office market has hit its maximum hype thanks to one property speculator from Queens and another who got his start in Brooklyn. Outsiders may not have higher average returns, but they have higher variance, so they dominate the tails of the distribution.

[2] For a while, WeWork’s head of IT was a literal high school student, nicknamed Joey Cables, who dropped out his junior year to focus on WeWork full-time.

[3] This is Paul Graham’s theory of unions, and also explains the existence of last-minute airfare, oil industry pay during boom times, and a handful of “strategic” acquisitions when an old-economy company buys out a pre-product startup. (The last case can go to even greater extremes; when the acquisition size is a round number, assume that the startup being acquired is a Veblen Good, and that the company is paying a billion dollars to tell investors “We’ve invested a billion dollars in…”)

[4] While this was big news ($, WSJ) at the time, it wasn’t news to WeWork employees or investors. Billion Dollar Loser cites multiple instances in which Neumann moved to a new office and requested the installation of a vent that would allow him to smoke there.

Further reading: I enjoyed Billion Dollar Loser, and recommend it as a great retelling of the WeWork story. I previously wrote about WeWork’s IPO prospectus here, and its aftermath here. And I wrote about Softbank and the “Capital Moat” strategy earlier.



Amazon’s Third-Party Data

Amazon has a rapidly-growing advertising business because of the company’s vast collection of first-party purchase data. Amazon knows what you bought, what you browsed for first, how long you spent on reviews—as long as that activity occurred on Amazon. To see what’s going on in the rest of the retail economy, though, they need to get data from users. Amazon is doing this through a shopper panel app, which gives users gift cards when they upload photos of receipts from non-Amazon purchases. There are a handful of independent companies that sell data on this, too, but the economics of doing it in-house are similar to those that affect alternative data in finance: it’s easier to amortize costs over multiple users, but a single user captures more upside, and has more control over exactly what data is collected.

The Lucas Critique in Finance

Cynically, published research on systematic investing strategies falls into roughly three categories:

  1. Results based on spurious correlations that never worked.

  2. Strategies that produce excess return, but only by taking additional risk, and

  3. Strategies that used to produce excess risk-adjusted returns, until somebody published them.

In a fun example of point #3, some papers a few years ago discovered that executives' language on conference calls could predict future returns. As a result:

Managers are emphasizing positivity and avoiding words or phrases known to be perceived by machines as negative. So it’s out with things like “claimants” and “cease” and in with the likes of “innovator” and “improving.”

One could take this a step further, and identify which CEOs studiously avoid using the entire set of harmful words while using as many of the good ones as possible. Hyper-compliance with a rule implies a desire to game it. It’s a bit similar to the old joke that American communist organizations in the 50s knew which of their members were undercover FBI agents, because they were the only ones who paid their dues on time.

Matching the Huawei Strategy

Last week, I mentioned a piece about how Huawei closes deals: even when the company itself isn’t getting subsidized, its customers are getting below-market loans to buy Huawei equipment. The US is trying the same approach, offering to lend Brazilian telecoms money to buy non-Huawei gear. An interesting feature of US/China competition is that both countries are trying to use their vast financial systems as a tool to get other countries into their respective orbits. China has a scale and coordination advantage, since its financial system is immense, its savings rate is high, and banks ultimately answer to the state. The US’s advantage is a sophisticated financial system; less firepower, but better aim.


Many market moves are notable because of what they imply about reality, but sometimes we get a natural experiment in how markets react to fiction. At a conference yesterday, short seller Jim Chanos said he’d “go long any of the space companies that have gone public because we know that space is infinite. There’s no price too high to pay.” In text, this looks like a plausible quote, but in context it’s obviously a joke—Chanos is a short-selling specialist, and a notably cynical one at that. But text is viral, and his quote was repeated online, causing shares of Virgin Galactic (ticker: SPCE) to briefly rise 7%.

This is telling, because it reveals what optimistic investors want to believe: that growth and a large addressable market are all that matters, and that the experts are refusing to acknowledge the obvious.

Demand and Logistics

In a normal recession, one of the forces that keeps the economy from contracting too much is the stability of government jobs. Government employees can get laid off in a recession, but their work is usually less sensitive to temporary economic ebbs and flows. That’s not happening this year: local government employment is down almost as much as overall employment, and the Oregon Office of Economic Analysis has a good post on why:

The bulk of the layoffs [in education] are tied to student workers, recreational centers, dorms, student unions, and the like… Besides education, the public sector does a lot of things. Employment here is down largely due to zoos, convention centers, recreation facilities, public pools, libraries and the like being limited during the pandemic. The losses in public administration are relatively small to date.

This is a reminder that the Covid recession is not like a normal recession; the shortfall in demand is partly driven by the fact that some kinds of demand simply won’t exist, at any income level, until the disease is not a risk. That implies a different policy mix than usual. Normally, a recession is a time to kickstart the economy to get people working again, but this round, policy needs to focus on making what The Economist calls “The 90% Economy” a harmless as possible, while waiting for the opportunity to get back to 100%.