The Economics of Noncompetes

Plus! Always a Bull Market Somewhere; SPACs; "Most-Hated Rally"; Pricing; Amazon and Seasonality

The Economics of Noncompetes

Some economists like to point out that one of the most valuable rights accorded to people and corporations is the right to be sued. It's a way to outsource your trustworthiness to the legal system, by making commitments that stay binding even if you'd prefer, after a while, that they didn't. But the right to be sued also creates the risk of spurious lawsuits.

Part of this is because noncompetes have a direct, positive-sum role, and an indirect and negative-sum one—and with many practices, the companies that get good at the positive-sum version may find that their best avenue for further increasing their economic advantage is to go negative-sum instead.

  1. The obvious reason for a noncompete is that it improves the risk-reward for training a new employee, and reduces the risk of having that employee walk off with critical proprietary information. Investment bankers will generally have a noncompete that lasts long enough for the deals they've worked on to mostly complete or get canceled, for example, which makes sense for all sides—it would be aggressive indeed for a bank representing one buyer in a competitive deal to poach someone who had worked for another buyer.1 This kind of restriction can apply, in a fuzzier way, to jobs where there's a network and reputation component. A junior banker's reputation is mostly a function of the logo on their business card, but as they get more senior it becomes more a function of who they are and what deals they've done (or avoided). Their employer can credibly claim that they wouldn't have the same personal cachet without the bank brand's initial cachet, but that argument gets harder to make as their individual performance gets better.

  2. Which brings up the other reason companies pursue noncompetes: it improves retention through the simple expedient of making it legally difficult to switch jobs. This is an especially strong force for specialized roles, where most of the jobs that use the relevant skills will be directly competing with the previous employer, making the noncompete more enforceable. This has led to longer and stricter noncompetes in quantitative finance, for example: "Six- to nine-month restrictions were once considered lengthy, especially compared to the three-month garden leave common at big banks and elsewhere on Wall Street. But that time frame has been eclipsed by 15- to 24-month sit-out periods, which have gone from rarity to commonplace over the last few years..." For systematic strategies, it's often more seamless to incorporate a signal into an existing system than it would be for more discretionary strategies—the trades are already happening, and if you treat a signal as a probability distribution of expected prices over time, you can basically add them together.2

Sometimes, a noncompete can serve more of a social function: it's a way for a jilted manager to spark a conversation with a recently departed employee, vent a little bit about how rude it was for them to quit, and then get calmed down by their lawyer if the noncompete doesn't apply, won't be enforceable, or just won't be worth the cost. It's the corporate version of sending a passive-aggressive text message to an ex, except that it generates billable hours.

Noncompetes can encourage workers who like their industry but dislike their company to find some lateral way to participate. If they can't build a particular business, they can build tools to build such businesses. Or they can build it somewhere else. David Neeleman actually did both: he sold an airline to Southwest, then got fired five months later, but was stuck with a five-year noncompete preventing him from working on a US airline. So he cofounded an airline tech company, which was eventually sold to Hewlett-Packard, and also helped start Canadian discount carrier WestJet. And then, once the noncompete expired, he was on to JetBlue.

There are ways to get similar incentive alignment without directly resorting to noncompetes. For example, vesting equity compensation, especially if it's back-end weighted, is essentially a way to financially penalize employees for quitting. More subtly, California's lack of enforceable noncompetes might encourage companies to use idiosyncratic technologies, offer unusually powerful developer tools, emphasize unique company cultures, and offer odd benefits—anything that makes the decision to leave a qualitative as well as financial one will make turnover less common. This is mostly not a deliberate choice: lots of those company-specific differences start with founders' personal preferences, not with some grand scheme to minimize turnover a decade later. (And anyway, early on more recruiting is through social networks, which puts a damper on turnover—which is also a way to improve employee retention, albeit one that's naturally hard to scale.) Even if it's not a deliberate choice, there's selection pressure that favors random workplace culture mutations that making quitting more difficult.

In finance, the classic way to reduce turnover, other than suing ex-employees, is deferring a chunk of their compensation. At funds, this is often a combination of punishing defection and rewarding loyalty; if I were required to invest half of my annual income in Renaissance's Medallion Fund with a multiyear lockup, I'm sure I'd find a way to cope with the imposition. But this adds a layer of operating leverage to the funds that do it: when the fund is going well, it's a bonus, but when the fund does poorly, it's a retroactive penalty. And it can be quite toxic when one part of the fund is responsible for the shortfall, or when one group is doing well and the rest of the fund is lagging. (As in many other domains, the fundamental attribution error applies; we have a moral call option where we feel that good outcomes are something we worked for and should get paid for, while bad ones were bad luck.)

Companies that can't directly enforce noncompetes will sometimes come to agreements, of varying levels of formality, about who they will or won't poach. These always read badly, but are sometimes the most direct way to lower the transaction cost of working with a potential competitor on some project. When Google was growing, one reason companies were reluctant to partner with them was that their best employees wanted to go to Google, which prompted some no-poaching agreements that eventually led to a substantial settlement. Widespread no-poaching agreements mean companies are coordinating to broadly keep wages low, but two-sided and role-specific ones looked more like a way for a company with desirable jobs to partner with a company that wasn't such a prize on a mutually-beneficial project. (If you read through the background of the no-poaching lawsuit, many of the interactions that precipitated these agreements didn't start with a discussion of how to keep salaries low. Instead, the usual email was from a CEO saying "We were working together, and your recruiters are cold-calling my employees. Rude!" Of course, the line between politeness and illegal anti-competitive behavior is a blurry one; an old-fashioned term for politeness is "being a gentleman," and a not-uncommon euphemism for antitrust violations is "a gentleman's agreement." It's also rude to undercut a company on price or compete with them on a partnership. )34

The fundamental problem for both employers and companies is that the default employment relationship in the US can end at pretty much any time, and for any reason. But there are mutual benefits to be had when companies take time to train employees, and when those employees take time to learn their company inside and out. Some countries try to solve this with policy—in Germany, Kurzarbeit is a setup where workers can accept shorter hours and get partial unemployment, so companies don't choose to fire workers in downturns and then rehire them later. Making it generally harder for companies to fire workers indirectly encourages training (and also makes it harder to quit; a company that's making a long-term commitment when it hires someone will be more judicious about that choice, so quitting workers end up with fewer places to go). Noncompetes themselves are a blunt instrument for achieving this, but people's desires and goals change all the time, and it's risky to invest in human capital when that capital can just walk off. They aren't optimal, but none of the available tools really are.


Always a Bull Market Somewhere

In general the IPO cycle for the lowest-quality companies follows the general financial cycle on a lag: they'll go public near the peak, either because they have an opportunity or because they were founded or reorganized specifically to target a quick IPO, or they'll go public during a market lull because new companies don't have conspicuously ugly charts and because banks have boom-time overhead and are more willing to lower their standards for revenue. So we end up with companies like AMTD Digital, which went public in July and peaked at a price 200x its IPO price. The company has minimal revenue, low float, and a CEO with a checkered background, who has been censured by Hong Kong's regulators ($, FT). And there are a few other cases like this. Magic Empire priced its IPO at $4 and traded up to $235.95 before collapsing. (It closed yesterday at $12.77.)

These are generally tiny companies, with few shares outstanding. They're quite expensive to short; the borrow for Magic Empire is 107%, and AMTD's was similarly high when there were shares available. So some of what's happening is the temporary market inefficiency where buyers have an easier time buying than sellers do selling. But what's also changed is that the feedback loop for retail investors is tighter. These aren't WallStreetBets stocks, but they are the kind of stocks that get promoted on Twitter, and on trader Discords. And there's a narrative beyond gambling: long owners mostly know that the companies aren't that fundamentally impressive, but they also know that when they buy, they're hurting short sellers, and that has its own appeal.


In other IPO news, SPAC issuance has reached a five-year low, with no issuance at all in July and only four small deals in August ($, WSJ). The SPAC boom always made a certain amount of unfortunate sense—retail investors like the speculative aspect, while institutional investors provided some liquidity through SPAC arbitrage strategies (the right to redeem for cash means that a SPAC at $10 is an option on a good deal whose premium is the cost of capital).5

One reason for the slowdown in issuance is that retail investors have largely moved on, but another reason is that while the supply of managers who are interested in launching a SPAC is large, it is not infinite, and the glut of deals means that many teams are tied up looking for transactions for entities they've already taken public.

"Most-Hated Rally"

Every rally gets described as a "most-hated rally," mostly by people who realize they should have been long earlier. The two hateable categories are a) times when the economy is doing well, and stocks are discounting that further and further into the future, and b) times when the economy is doing poorly, but the market is already betting that the recovery will happen. (The S&P hit its post-crisis low in March of 2009, while unemployment peaked 1.3 points higher in October of that year). The last few months' rally is a controversial one because it's hard to decide which reason to hate it for: is the bet that the economy is going to quickly recover from a short recession, or that there really wasn't a recession at all? Either way, equity hedge funds are, relatively speaking, sitting out the rally, with record negative bets on S&P futures. Which partly implies that they're bearish, but also implies that long/short funds see a lot more value to add with stock selection on the long side. Shorting is always hard, but for investors who've done well paying nosebleed multiples for growth stocks, it's especially hard to decide what constitutes a too-expensive multiple—whereas modeling growth a few years out and putting a multiple on that is a skillset they still have.


Delivery company Delivery Hero talked up its "million pricing decisions per month" on a recent earnings call, as part of an effort to reach profitability by the back half of this year. One model of the food delivery business is that, even in the absence of convenience, it's possible to earn a premium simply from better pricing decisions—the cost of delivery is partly a fixed cost for an opportunity to play more transaction-size-maximizing games than a typical restaurant can play.

Amazon and Seasonality

Amazon is raising seller fees marginally over the holidays. Some of this is purely passing along inflation, but, as with their plan for a second Prime Day, one of the drivers is an effort to reduce seasonality. Given the cost of recruiting new warehouse workers and the difficulty of retaining them, anything that can spread out the shopping season, even marginally, pays off.

Disclosure: Long AMZN

  1. Though this kind of thing can happen! California is famously reluctant to enforce noncompetes, and there have been cases where two tech companies were in the running for the same deal, and one of them hired someone who was working on the deal for another one.

  2. Discretionary investing is different partly because it's less repeatable—the repeatable parts are all Python scripts now—and because there's a high cost to communicating them and persuading someone to pursue them. Usually the de facto arrangement for discretionary research is that one analyst pitches a complete idea, with company-specific and industry analysis. In a discretionary strategy where some analysts focus on the macro picture, some predict company cash flows, and others figure out where in the capital structure to make the trade, you'd expect ideas to be more portable and noncompetes to be stricter.

  3. One thing unenforceable noncompetes do is to encourage company specialization. If there are ten full-stack companies in an industry, each one can hire from all the others. But if the industry is instead a stack of ten specialized businesses, each dominating one layer of the industry, then many skills don't transfer as well.

  4. Another thing to note is that many of these agreements were made years ago, when the companies involved were smaller. Right now they read like super-profitable trillion-dollar businesses conspiring to suppress wages, but in fairness they were merely fairly-profitable multibillion-dollar companies doing this. There comes a point in a successful company's existence when it's small enough to get away with some dubious behavior, but powerful enough to get a lot of leverage out of such behavior. This can be more visible within the company than outside of it; Google knew how much wealth it was creating, and how much it would profit from capturing that wealth later on, but there were plenty of skeptics about Google-the-company and Google-the-stock in 2006. A company trading at over 100x earnings in an industry that had a high casualty count reminded plenty of people of then-recent bubble excesses.

    It’s almost a rite of passage for companies at the steepest part of the S-curve to have a little break from trying to take over the world during which they deal with the PR and regulatory blowback from behaviors that struck them as scrappy at the time but have been publicized at a time when they look more threatening.

  5. SPAC warrants provide another fun kind of liquidity; they're a fun call option whose delta can go negative under certain circumstances. When a SPAC with an announced deal is trading above its redemption value, fewer holders will redeem. While redemption is bad for the company, it can be especially bad for short sellers, because a sudden decrease in shares outstanding means they'll have to scramble to acquire shares—the higher the short interest pre-close, the more hated the deal and the more likely redemptions will be high, too. So if a SPAC with an announced deal is trading above $10, and the price drifts down, the warrants are worth more since there are higher odds of a volatile price spike soon.

    Complex financial structures are essentially a way to subsidize liquidity provision from institutional investors by giving them more arbitrage opportunities.