Welcome to the Friday edition of The Diff! This newsletter goes out to 26,518 subscribers, up 702 from last week. In this issue:
- Keeping Options Open: The Equity Compensation Story
- Prime Elasticity
- Labor Arbitrage
- Location Arbitrage
- Legal Accessibility
Keeping Options Open: The Equity Compensation Story
This piece is also featured on The Archive, from Compound.
A perennially tricky problem in compensation is the principal-agent issue: at a given company there are outside investors who provide capital and the managers and employees who are stewards of it, and their goals don't always align. The history of compensation is a long history of trying to get them to think about things in the same way, so the company wastes as little effort as possible on zero- or negative-sum wealth transfers and spends as much time as possible on wealth maximization.
The usual way this is framed is in terms of getting employees to think like owners.1 In an ideal world, owners, and employees would have similar economics such that there isn't a direct conflict between them, but there are three issues with this:
The investors, by definition, are the ones with the capital, and they usually can't insist on only backing founders who co-invest with them.
If they give founders and employees some form of asymmetric upside, they create a scenario where some share of increased value goes to one party, but all of the decrease in value accrues to the other side.2
On the other hand, employees and founders have asymmetric downside, since they're less diversified than their investors. It's a lot easier to create a portfolio of 20 angel investments than to create a portfolio of 20 angel-funded jobs.
Balancing points 2 and 3 is possible in theory, but financial history is riddled with attempts to do it right with varying degrees of success. Like most other phenomena, employee equity and stock options are much older than one might think: the British East India Company had a sort of reverse equity compensation scheme: shareholders could vote for directors, and the company's directors controlled hiring for jobs, so shareholders were incentivised to vote in favor of directors who could promise them jobs later on. This created unusual incentives for both Directors and shareholders.3
Meanwhile, the Dutch East India Company led to the creation of a vibrant market in shares, futures, and derivatives; Joseph de la Vega's Confusion de Confusiones has a section that not only discusses trading options but also delta-hedging his exposure when he changes his mind. Even in 17th-century Amsterdam, there are traders who don't know the theory all that well but are experts on the practice.
The worlds of tradable options and equity compensation took a long time to converge. A watershed event accelerating this occured in 1902, when the newly-organized US Steel created a "Stock Subscription and Profit Sharing Plan," which gave employees the opportunity to buy shares at below-market prices, to defer payment for up to three years, and to get a bonus if they held for five years.4 By the 1920s, over 70,000 US Steel employees took advantage of the plan each year. Being able to borrow money at no interest on a non-recourse basis to buy stock is functionally equivalent to owning a call option struck at the price of the stock—the holder gets all of the upside but doesn't face the downside.
The eventual combination of stock options as a tradable asset and equity-based compensation as a way to motivate employees finally came together in a more permanent way in the 1950s, when part of the motivation in question was taxes. High tax rates create all sorts of surprising incentives, ranging from efforts to lose money on paper with oil and cattle, both of which had associated tax benefits; to paying employees in fringe benefits rather than salary (executive dining rooms, company cars, and the three-martini lunch are all artifacts of this) to; more recently, trying to frequently realize losses on crypto transactions because wash-sale rules don't apply ($, WSJ).
And, in the 1950s, one of the tools available to companies was the stock option: following World War Two, marginal taxes rates for executives were very high, with most paying between 80-90%. The Revenue Act (1950) introduced ‘restricted’ stock options (later called ‘qualified’ or ‘incentive’ options). This new option was subject to capital gains tax, at the much lower rate of 25%, and the tax was deferred until the disposal of the stock. Before this, stock options had been taxed as income, and so there were no advantages to receiving them as part of compensation. In 1950, almost no executives in the sample tracked by Historical Trends in Executive Compensation were granted stock options, but in 1951, 18% of executives adopted restricted stock options as part of their compensation.
Stock options were getting more common outside of the executive suite in the 1960s. The defense conglomerate Litton used options, so when Henry Singleton left that company to found Teledyne, he brought the practice along. Ling-Temco-Vought, which acquired companies and then spun off tracking stocks for their more interesting divisions, would compensate the heads of those subsidiaries with stock options on their own company instead of the entire firm. In The Money Game, written in 1968, the author complains that it's hard to get accurate gossip from talking to engineers at tech companies, because they all own options (so they want to hype their own company) and because they're also just as market-obsessed as the investors pumping them for ideas.
One reason options are such a key part of tech DNA is that they were a big factor in the birth of Intel, from which many modern management norms are derived. Intel was getting disrupted, and doing the disrupting, before it was cool. And before it was Intel, the founding team worked at Fairchild Semiconductor, where they were shareholders, but where the parent company, Fairchild Camera, had an option on their equity. The Fairchild Semiconductor team wanted to grant employees options; Fairchild Camera said no, and soon enough the Fairchild Semiconductor team was a bit smaller as Gordon Moore, Robert Noyce, Andrew Grove, and others left for greener and more incentive-aligned pastures.
When companies go public early in their lives, it's relatively easy to decide where options should be priced. There's a market price for the stock, so there's a natural strike price for the options, and better tax treatment for options granted with a strike price equal to the market price. And its volatility is measurable, so it's easy to apply Black-Scholes and get a valuation. But for private companies, pricing options involves some guesswork. And it used to be lucrative guesswork: underpriced options a) were a cheap form of compensation, and b) meant that anyone investing using a tax-free but capital-constrained vehicle could use it to acquire some very affordable options.
Options used to be valued through rules of thumb; this piece says the usual rule was that common stock was worth 1/10th of the valuation of the preferred stock.5 For basically unrelated reasons—Enron managers preemptively executing stock options to cash out during the company's collapse—the rulebook on stock options was totally rewritten in the American Jobs Creation Act (2004), adding rule 409(a) which required that companies make some effort to value options rather than use a rule of thumb. (The rule of thumb turns out to be a little off, but at least to have a directional correlation: 409(a) valuations tend to be 30-40% of preferred valuations.)
This has created a small industry of firms that perform valuations for early-stage companies so they don't accidentally trigger tax penalties. It ends up functioning as, essentially, insurance: pay a few thousand dollars each year in order to avoid paying a much larger penalty later. 409(a) valuations are still fairly subjective, especially in the earliest stages, and there's room for management to push the values down. Subjective features like how long it will be before there's a liquid market for the stock, which public companies get included in comparable valuations, or what metric they're compared to can work. (How much do metrics matter? A money-burning startup can't use an enterprise value to EBITDA ratio, but a SaaS company that started selling its product mid-year will have a much higher annual recurring revenue number than its sales, and even after that point sales will lag ARR if it's growing fast.)
Part of the point of setting a low value on options when it's possible is that "options motivate employees" is a claim that works at two levels of abstraction. The obvious one is the incentive issue outlined earlier in this piece: the more someone's economic rewards approximate the wealth they create, the more their interests are aligned with shareholders. But the other form of motivation is that it's always exciting to see a number go up, even if it doesn't lead to large economic rewards. So the incentive for low strike prices is not just that they marginally affect the payoff from options, but that every subsequent valuation can go a little higher by closing the gap between the 409(a) valuation and the real value of the options.
Momentum is a signal that investors and employers respond to; it produces some excess returns in public companies, and it affects behavior in private ones, too. So the basic logic behind under-pricing options remains the same even as the details change. It's a lot easier to keep people motivated when they own something that used to be cheap and has gone up a lot since then—witness the élan of the laser-eyes crowd—and that's easier to accomplish when there's some room to set the price.
Diff Jobs is our recruiting service that matches Diff readers to interesting roles at companies looking to hire them. Some positions we're working on right now:
- Sales/business development roles in multiple industries, including crypto (US, remote).
- An technically skilled execution trader (think more Python than Excel) at a fintech company working on new systematic strategies (NYC).
- A business analyst with Looker experience at an ecommerce platform going through explosive growth (US, remote).
- A head of security research at a crypto firm (US, remote).
- A senior backend engineer at a company that uses machine learning to stop fraud. (Remote, Central Europe Time ideal.)
Amazon Prime probably has 150 million US subscribers and 98% annual retention, and Amazon may increase that price. Prime was an early step along the path to making Amazon's economics a) good, and b) inscrutable. It makes it easier for people to instinctively order products from Amazon, and while it raised overall shipping costs, it lowered the marginal cost of making shipping faster, which further increased order frequency. Meanwhile, an increasing share of the value Amazon gets from shopping shows up as ads rather than fees on sales, since the ad prices are determined by what profit sellers expect to earn rather than by Amazon's best guess at the ideal fee schedule. Estimating the churn hit from a Prime price hike is one of the more high-stakes pricing decisions out there; it is valuable, and gets more so as people do more of their spending on Amazon, but it's incredibly hard to determine what price would make people switch to cheaper membership programs. And that's especially difficult because whatever price that is is also the price where churn would happen over time, which would have second-order effects on how merchants think about Amazon, and how further investments in logistics and media pay off.
Disclosure: I am long AMZN.
In most parts of the US economy, unions have gotten weaker over time; they don't have the same legal leg up they used to, and companies that successfully offshored production to places with weaker unions were largely able to undercut the companies that didn't. Airlines remain an exception to this, in part because the workers are a complement to expensive equipment that has fixed costs whether or not it's in use.6 A new airline from JetBlue founder David Neeleman has found an alternative, by using visas to attract pilots from countries where travel is more impacted by Covid. Part of what keeps airlines unionized is that negotiations are not a one-shot game: unions have learned that they really don't want to drive companies into bankruptcy, and companies have learned that getting a temporary edge is not worth the cost in future negotiations.
Northwest Arkansa is offering remote workers $10,000 to move there. It's worth thinking about the effects here. If someone can live anywhere, and decides based on cost and cost-of-living, then getting cash upfront for moving is a great way to make the decision easy—once you're looking outside of big cities, there's a lot of America that's incredibly cheap, and cash upfront is a good way to break a thousand-way tie. The hope might be that these workers will create a tech cluster, but $10,000 has another selection effect: it targets people who don't think that their current location has $10,000 worth of socializing, entertainment, or job opportunities—i.e. people who are relocating physically but plan to stay put in terms of their career. It can still pay off, of course; sales and income taxes add up. But it's a more purely financial transaction, where the state is underwriting the risk that people who moved in response to a one-time incentive will move again when the incentives shift.
Via Marginal Revolution.
Lawmakers have proposed new rules requiring companies to explain their terms of service in plain English. This has two effects:
- A small percentage of the time, it prevents scammy companies from doing nearly-fraudulent models where they sell samples of a product for a low cost and then sign buyers up for an auto-renewing subscription.
- Much more often, they get customers used to scrolling past basically meaningless legal agreements.
It's less about what customers agree to—aside from the literal scams, these usually don't have a big effect on outcomes. Instead, it's about the ongoing uncertainty: if every company is required to display its terms plainly, then companies won't face the market-for-lemons problem that complicated terms make it look like they have something to hide.
Though it's worth noting that founder-friendly term sheets and a commitment to backing founders both try to solve the problem in the opposite direction: getting owners to think like employees, or at least eliminating a set of possible scenarios where their interests can diverge. ↩
This is one explanation for the sometimes sharp-elbowed corporate culture in some investment companies: an analyst's upside on a good recommendation is multiples of their salary, and their downside is that they need to find another job. But if they work at a sufficiently prestigious firm, it's fairly easy for them to find another job somewhere else. So the implicit goal of screaming, throwing phones, engaging in weird mind games, etc., is to give someone who could earn a $500k bonus from a good idea roughly $500k worth of emotional downside from a bad one. So the natural tendency is that the better an investment company's brand name, the more likely it is that either a) they have high turnover and a pretty aggressive culture, or b) they have low turnover and lots of deferred compensation. ↩
In this reverse equity system, there’s a chance that you could trade-off between your salary and your equity. Imagine yourself as a Director seeking election, you might solicit the backing of shareholders, who you will give middle manager jobs to in return, in order to gain election. This will happen despite the fact that they would make poor managers. The poor managers would (theoretically) impact negatively on the share price, but in return you’ve gained a Director’s salary, and so individually you’re better off. However, the Company safeguarded against self-interested short-termism by holding elections for all 24 Director posts annually. Furthermore, after serving four consecutive years, Directors would have to take a year off. A Director seeking re-election after a year off, would have to rebuild their support base: an easy task for someone who positively impacted the benefit of everyone else’s shares, but far more difficult for the Director with a reputation for giving patronage to incompetent managers. The intended effect is a system that self-corrects against self-interest; favouring instead the Company’s long-term collective good. This term limit system makes sense in a context where it’s hard to perfectly align incentives—the trade of losing good managers periodically in exchange for shaking out corrupt ones makes sense. While it’s unlikely to be a business school case study, the British EIC might be able to teach modernity a thing or two about corporate governance. In fact, in one particular category companies do use a similar technique: traders at investment banks are sometimes required to take vacations, because it’s a chance for someone else to sit in their seat and ensure that they’re not hiding losses. ↩
This was partially a way to align employee incentives with investors', but had another important angle: US Steel was formed through what was essentially a series of leveraged buyouts of steel companies, so its major shareholders had large, levered positions in the stock and were eager to sell. The five-year bonus plan was partly a way to lock up shareholders so there would be less net selling pressure. ↩
A rather low number! Since common stock is junior to preferred, it functions similarly to an option, but startup outcomes are pretty binary, so the common stock will either be worthless or have a value that converges on the value of the preferred, at least the early rounds. As Chris Dixon pointed out a while ago, if you plug a startup-appropriate level of volatility into the Black-Scholes model, you get options that are basically worth as much as the underlying assets, and the same math applies to preferred. The intuition here is that liquidity preferences are basically the "strike price" at which common stock is in the money. Take two companies: one has 10m common shares issued for $1 apiece, and 10m options with a strike price of $1. Up to a $10m exit, the common stock gets 100% of the returns, after which they're split 50/50. Now consider a company with $10m in preferred with a 1x liquidation, and 10m common; the payoff functions are the same. So options on common stock in a company with liquidation preferences are, in a sense, an option on an option, although the second layer of optionality doesn't matter for economic purposes so much as for the taxes. ↩
One argument on unions is that they've been replaced by occupational licensing, which has some of the same function—decreasing labor supply to increase its price—while having better legal protections by definition. Pilots are the sort of job where licensing makes a plenty of sense, but perhaps unions have remained strong because they've been better at controlling supply than licensing could hope to be. ↩