Netflix CEO Reed Hastings is celebrating the principle that great software programmers are orders of magnitude more productive than average programmers. The implication is that sky-high salaries for these rock stars are worth it.
Now, it may very well be the case that the best programmers are orders of magnitude better than average programmers. I’ve seen a similar thing on display during examinations for gifted students: inevitably one student finishes the exam in half the time and walks out with a perfect score, while the rest of the gifted struggle on.
Just how many orders of magnitude smarter is that student, relative not just to the other gifted students in the room, but to the average student who is not in room?
But while the rock-star principle may justify the high willingness of Silicon Valley firms to pay for talent — the more value an employee brings to a firm the more the firm can afford to pay the employee and still end up ahead — that doesn’t mean that as an economic matter a firm must pay rock-star employees higher salaries.
Far from it.
Economic efficiency requires that great programmers be put to use programming, otherwise society loses the benefit of their talents. But the minimum salary that, as an economic matter, a tech firm must pay a rock-star programmer to induce the programmer to program is just a penny more than what the programmer would earn doing the programmer’s next-most productive activity.
If the programmer isn’t good at anything but programming, that number might be $15.01 — the $15 minimum wage Amazon pays its fulfillment center workers plus a penny — or even something lower, as the programmers I know would have a tough time sprinting around a warehouse all day.
A programmer might be worth $100 million as a programmer, for example, because the programmer is capable of delivering that much value to software. But to make sure this person actually delivers that value, the market does not need actually to pay the programmer $100 million, or anything near to that amount. All the market needs to pay the programmer is a penny more than what the programmer would earn by not programming.
And if rock-star programmers tend only to be rock stars at programming, as I suspect is the case, that number might be pretty small, indeed, on the order of what average programmers make — if not $15 an hour, which is a bit of an exaggeration — because the rock-star programmer is likely to be average at programming-adjacent pursuits.
If the most the programmer would make teaching math, playing competitive chess, or just programming for non-tech companies that will never earn the profits needed to pay rock-star salaries, no matter how talented their employees, is a hundred thousand a year, then that plus a penny is all that economics requires that the programmer be paid for doing programming. Not $100 million.
So why are rock-star programmers earning the big bucks in Silicon Valley? Because tech firms compete for them, bidding up the price of their services.
Tech firms know this, of course, and once tried to put a lid on the bidding war, by entering into no-poach agreements pursuant to which they promised not to try to lure away each others’ programmers by offering them more money.
There is no reason to think that these no-poach agreements were inefficient. Unless you believe that programmers can contribute more to some tech firms than to others, in which case the bidding wars that drive rock-star compensation sky high are allocating programmers to their most productive uses. But that seems unlikely: does making Google better contribute more to America than making Amazon better?
(The agreements also could not have created any deadweight loss, because perfect price discrimination is the norm in hiring programming talent: firms negotiate compensation individually with each programmer.)
All the no-poach agreements did was to change the distribution of wealth: limiting the share of a firm’s revenues that programmers can take for themselves.
Indeed, the no-poach agreements probably contributed a bit to the deconcentration of wealth.
A dollar of revenue paid out to a smart programmer goes in full to the programmer, whereas that same dollar, if not paid to the programmer but instead paid out as profits to shareholders, is divided multiple ways between the firm’s owners. Competitive bidding for rock-star programmer salaries concentrates wealth, and the no-poach agreements spread it — admittedly to shareholders, who tend to be wealthy, but at least the dollar is spread.
The antitrust laws intervened just in time, however, to dissolve these agreements and punish Silicon Valley firms for doing their part to slow the increase in the wealth gap in America.
Today’s antitrust movement has argued that antitrust should break up the tech giants in part to prevent them from artificially depressing the wages they pay the little guy. I’ve argued that would be a mistake, because breakup could damage the companies, reducing the value they deliver to society and harming everyone. Regulating wages directly is a better idea.
But you don’t just make compensation fair by raising low wages. You also have to reduce excessive wages. One way to start is just by allowing the tech firms to conspire against their rock stars.
And once tech firms have finished conspiring against their overpaid programmers, they can start conspiring against another group of employees that is even more grossly overpaid per dollar of value added: their CEOs.
Well, that we might have to do for them.