The labor market is like the pricing page for a certain category of SaaS product. For legible everyday needs, there's usually a pretty standard price list and a very standard bundle of benefits. Terms get more complex for both sides as the price goes up. And, past a certain point of complexity, the labor market says the equivalent of "Call for a consultation and we'll see if it's appropriate to give you a quote."

There are a few axes of complexity here. The simplest is that for some jobs, workers are price-takers—if a job requires few specific skills (fast food, retail, ride-sharing), it competes with all of the other jobs that require that skillset. There's still some variation; a 9-5 shift is easier to hire for than a graveyard shift, the wage differential between making fast food and delivering it depends on what premium workers put on setting their own schedule, etc. But for the most part, it's a big labor market where employers have a vested interest in making workers as undifferentiated as possible.

Oddly enough, this kind of hiring market also applies to skilled workers who go through a standardized training regimen before they enter the labor market. As far as labor's bargaining power is concerned, a freshly-minted Harvard JD has as much negotiating leverage with Latham & Watkins as a new high school graduate has with Walmart.

But if the job is in sales, a new dynamic kicks in. It's very straightforward to measure a salesperson's productivity, at least most of the time: just count up the revenue they generate. There can be some variance within this, depending on the product—if there's room for price discrimination, there's also a need for guardrails around discounting, lest one aggressive salesperson kill off a company's pricing power. But in general, a sales team can set their compensation so that pay is competitive for the producers and the weaker salespeople can't really afford to stick around.

But even that's just a rough description of how salespeople get compensated. The mechanics of paying someone a percentage of revenue from contracts they sign, perhaps with cutoffs, premiums, etc., are part of the formula, but the other piece is who gets access to which potential customers. There are some products that get sold to, and are worth selling to, a range of customers whose size varies by orders of magnitude. In that situation, being the salesperson in charge of closing a few Fortune 50 deals versus being yet another person who spends all day in Zoom calls with small business owners has a bigger economic impact than the exact choice of commission structure.

For less measurable jobs, there's another easy hack for aligning incentives: just pay people partly in equity. This is a noisy approach, but has some appealing characteristics: it means that they're paid based on how the whole company does, which reduces their incentive to benefit their team at someone else's expense. And it means that everyone has at least some financial incentive to police one another's behavior. It also fixes an incentive-misalignment problem: founders and investors might want to sell their company, but for the average employee that means risking layoffs. But if that average employee can also cash out with much more than they'd lose from being out of work for a bit, it softens the blow. This effect is a lot stronger early on, when people have points of equity rather than fractions of a basis point, but the tradition of equity compensation tends to persist long after the point where the average person getting equity compensation has the means to influence the stock price enough to matter financially.

In some jobs, there's a natural way to capture exact P&L responsibility. If you run a Chipotle, or a pod at a hedge fund that trades restaurants, the business you manage will produce a nice, clean profit and loss number at the end of every year. And it seems incentive-aligned to just pay you a portion of it. After all, that's the logic behind funds charging carried interest in the first place: 20% is equivalent to splitting the equity in a business 80/20 in favor of the investor and giving them a 1x liquidation preference.

But in both cases, the model still requires some guardrails: Chipotle doesn't want to reward managers who bump gross margin up a few basis points by replacing some of the rice with sawdust, and a hedge fund doesn't want to pay for the equivalent (writing out-of-the-money puts or taking equivalently asymmetric-downside risks).

In general, paying someone a standardized cut of the profits they produce, without many guidelines, is equivalent to arguing that every stock ought to trade at the same P/E ratio. An ideal compensation scheme would reward someone for longer-term behavior and penalize decisions that borrow future profits at a ruinous interest rate. But actually putting that into practice is painful, so instead the goal is to minimize manager discretion for any decision that could have this kind of payoff, in order to isolate the results of pure skill.

But that ends up creating a paradoxical problem: when talent is legible and measurable, and it directly predicts profits, suddenly those workers have all the bargaining power! They know exactly what they're worth, and employers are competing mostly on the basis of how small a cut of the upside they're willing to take.

But there are solutions to that, too; when workers have fungible skills that could be deployed at multiple companies doing the same kind of work, their employers want to optimize compensation around retention—paying them more every year is perfectly fine if it means they'll actually stick around year after year, over which time they'll presumably get better at their job and better at working with their colleagues. So liquid markets in legible talent lead to illiquidity and opacity: hedge funds want it to be expensive for their best portfolio managers to quit, and they also want these managers to always wonder how much money the fund is really making from them. Big funds will analyze the trades portfolio managers make, come up with a model for which ones are the best ones, and size them up (or hedge out likely duds). So even if someone knows their P&L payout to the tenth of a percentage point, they don't know whether that P&L is 100% of the value they provide to their employer, or a small fraction of it. Long noncompetes, deferred compensation, and clawback provisions all add some artificial transaction costs to the decision to leave, with the ultimate goal being that the best-paid people don't quite know what they're worth, and that it's expensive for them to find out.

Complicated and opaque pay structures affect a tiny fraction of the workforce, because most jobs' outputs are harder to measure and because most of the time, it isn't worth it. Sometimes, they seem excessively elaborate. But hiring a fund manager who produces $50m/year in alpha is basically acquiring a business that produces $50m in EBITDA, and a merger at that scale probably requires a lengthy and detailed contract and a lot of give-and-take around incentives. As the amount of data thrown off by workers rises, and as tools for analyzing data get better, you should expect these norms to work their way down to the rest of the labor market. It'll be a while before we live in a corporate panopticon where Starbucks is monitoring the microexpressions of every customer in order to provide a retention bonus to the most personable baristas. But that's the direction to bet on.

The varieties of compensation models come up often in The Diff. Some highlights:

• The economics of alpha capture ($), and AI deployment as alpha capture for everyone ($).

• You and your investment research.

• What’s happening in tech compensation ($).

• The history of options as a compensation tool.

Join the discussion!