The economics of developer pay

Spoiler alert: there is virtually no difference in the mechanics of work done between $100 an hour, $200 an hour, and $30k a week — all of the leveling up there is in sophistication on who you go after, what engagements you propose and deliver, and how you package things for clients. — Patrick McKenzie, “Talking About Money”

Software developer pay is a fraught topic. It’s treated as a state secret even among friends, discussed privately, in whispers. People throw rocks at the Google Bus anyway. I have no idea what my friends earn, but everyone “knows” that anyone “in tech” makes tons of money. Apparently, they also hate the homeless, and run over kittens in their Teslas, just for fun.

I’m exaggerating, but only just.

Bay Area developers are a well-paid bunch; that’s for sure. But after a decade of recon, what surprises me isn’t the absolute amounts — lower than surgeons, lawyers, and finance jobs, mind you — but its variance. In San Francisco, where a 1-bedroom apartment sets you back $3,000/month, and taxes run 40–45% all-in, young companies are hiring debt-laden new grads for $90k. That’s not a bad income, but it’s peanuts compared to the $150–175k/year or more (counting stock) new grads earn at facebook. And the facebook offer only gets better over time, as promotions, stacked stock grants, and bonuses roll in.

Other industries don’t have differences like these. If you’re making $12/hr working the front desk at a hotel, 2–3x across town is the punchline to a joke, not something that actually happens. Further up the income scale, my wife is an architect, and it doesn’t feel like differences of more than 20–30% happen that much. I’ve been chewing on this little mystery for a while, and I think I’ve finally figured it out.

Developer pay is set by two forces: productivity, and what negotiation theorists call “best alternative to negotiated agreement”, or BATNA for short. I’m going to explain how these work in the industry, and offer a few anecdotes I feel validate the model.

Labor productivity

The first factor is what economists call “labor productivity”.

The common use of “productivity” is, “how much a person got done”. A person who digs one hole per hour is half as productive as a person who digs two. Fine. But people do a lot of things: dig holes, make ice cream cones, manufacture drugs. How should we compare across activities? The question seems a little ridiculous, but that never stopped an economist.

The straightforward, if imperfect, solution to this dilemma, is to define “productivity” as “$ produced/employee”. So an economist would say a firm earning $1 million/year with ten employees is half as productive as the same firm with only five. It doesn’t matter what kind of activity the two firms are doing; one might be a restaurant, the other, an auto garage. Doesn’t matter. We just know that, in terms of “labor productivity”, one firm is 2x the other.

Labor productivity is a really interesting metric because it’s dead-simple to measure, yet, entire careers have been made arguing over why it behaves how it does. It’s easy to see why. Let your mind wander:

• Are large companies, in general, more productive than smaller ones? (Economists tend to think the answer is “yes”)
• Do management techniques, education, or other “people factors” influence productivity?
• Is it fair to compare low-margin industries, like restaurants, to high-margin ones like semiconductor manufacturing, or software?
• Why do some economies (nations) have higher/lower productivity?
• How does prior or future investment in tools, processes, education, etc. affect labor productivity?

One thing economists are pretty sure of: high-productivity industries and firms, tend to pay better than low-productivity ones. It’s not that big of a leap: a company earning more $/employee has more “capacity” for higher wages. In fact, in tech, many of the most profitable companies pay the most even as they earn the most profit. Think about it: a company that pays more, even while earning more profit for its shareholders. If that isn’t magic, I don’t know what is.

The “scaled platform operators” — facebook, Amazon, Google, Netflix — earn tons of revenue without many employees, the very definition of high productivity. Take Apple: $34 billion in gross profit last quarter; worldwide operations, but most design work is done here in the Bay Area, with only only 25,000 employees. That’s $1.360 million/employee. Any surprise they pay near the top of the scale? Ditto for hedge funds, top law firms, and other “big money, small staff” operations.

To be clear, I’m not claiming Apple is moral, or good, or anything else. Just that they (a) earn a lot of money (b) using a small number of employees, which means they’re going to pay a lot.

But that’s only half the story. High-productivity firms can pay a lot — they have a high ceiling — but that doesn’t mean they will. Nobody wants to overpay. All else equal, ownership (“Wall Street”) would rather keep more money in the till, than pay it out as wages. Enter negotiation. Who needs the other more?

You’ve got options

When two parties strike a deal, whether a job, house purchase, or even plans for a date, they’ve reached what theorists call “negotiated agreement”. Bargaining power — the ability to get one’s way — will tilt toward whoever has the best alternative to negotiated agreement: BATNA.

When a company wants to hire someone, who has the upper hand? It depends on a few factors:

• Time: does the employee need a job quickly? Does the company need to fill a key role?
• Rare personal attributes: is there something unique or special the company needs? Credentials, reputation/relationships, knowledge, know-how
• How many companies are hiring for this role?
• Does the company have a strong “employment brand”? Do they have a strong pipeline of inbound interest/applicants, giving them access to more inbound candidate flow?

Empirically, in tech, the highest-paid people seem to be those with (a) a personal skill or quality that’s hard to acquire (takes a long time, costs a lot) with (b) outsized perceived influence on the business/strategy of (c) a high-productivity firm. Those on the low end of the ladder are relatively interchangeable, and tend to work in lower-productivity firms.

Pay isn’t everything, but it’s worth considering. Some personal anecdotes follow.

Validating the theory: some observations

My most highly paid friend is a silicon designer at Apple. He has rare skills (a PhD and strong experience), and works at a very high-productivity firm.

Another friend of mine worked in finance for a while and then took his network to a financial technology startup. They needed him to help raise capital for a lending product. I don’t know what he makes, but I do know that in a company where security is critical and everyone uses Macs, they support Windows just to keep him happy. You do the math.

Startups don’t seem to pay as well as larger companies. You can’t blame them; they don’t have the cash, though they hand out stock. A lot of them get middling talent. The best don’t bother interviewing at no-name companies with non-famous founders; they expect to be wined, dined, and recruited. I’m coming to think recruiting is the number one, most important skill for a tech entrepreneur.

I’ve spent the last decade toiling at medium-productivity firms as a generalist developer. I think I’m good at my job and learn quickly, but I haven’t worked at particularly high-productivity firms, nor do I have rare skills.

Consulting doesn’t pay as well as product companies, at least, not successful ones. Again, productivity.

All big companies start as low-productivity small ones. Is it fair that later employees reap the rewards of earlier “low-productivity” efforts? Handing out stock can partially mitigate this. I still think everyone, from the engineers up to the CEO, of big, late-stage companies are overpaid relative to the employees and investors who took real risks at the earliest stages of the company. Nothing about the situation will change while the companies are posting record profits every quarter; “if it ain’t broke, don’t fix it”.

Final thought: ignore what’s hot. Naval Ravikant says you should follow your own intellectual curiosity, and he’s right. By the time something is “hot”, it’s too late to become a ten-year expert in it. If you want to be good enough to lead strategic initiatives at Google, or be a startup founder, you need to be five years into something before the rest of the world realizes it’s even worth doing.