$3.9B is only their profit for the quarter. Over the last year, they've reported a little over $13.1B in profit (>$600k per headcount).

That's what I was missing!

Well, one thing is that the marginal profit of labour is bound to be diminishing at such scale. However, keep in mind that it's not marginal revenue, but profit. This means that it's still a viable strategy to increase the amount of employees for them. That drop is not necessarily an issue in accounting terms.

Why is that interesting?

Companies should hire more people until marginal profit per new hire is $0 to maximize profits. And that number says nothing about marginal profits.

No, because hiring affects future income and expenses, not just present.

Companies don't try to maximize total profit, but rather return on invested capital.

This can't possibly be right. Why would you forego making extra money to keep your statistical average up? What are you getting from the average number?

Your theory predicts that price discrimination can't happen because companies would rather sell M widgets at X price than sell M widgets at X price and N widgets at Y price.

It's completely right.

Why would you invest $250K for a year's pay of an employee for $0 return. You can invest that $250K in a CD for 2% and get $5,000.

Shareholders are going to have no interest (long term) having companies invest their capital in the form of cash on the balance sheet at below market rates. Hence dividends or buybacks if they can't demonstrate sufficient roic on internal projects.

You can back into a target revenue/employee by looking at risk free rate of return + equity premium (say 7%), net margin (35% for FB this Q), and say if an avg employee earns $250K .... then they'd want a bit less than $800K/employee (35/100 * x >= 250k + (.07 * 250k)).

I don't think your comparison to widgets holds up. Would a company invest capital in more production capacity to sell more widgets at zero margin (all else being equal)? No, they'd distribute it to shareholders so they could invest it in bonds or other companies, and that's the analogy.

> I don't think your comparison to widgets holds up. Would a company invest capital in more production capacity to sell more widgets at zero margin (all else being equal)? No

Who said anything about zero margin? You're claiming that if I can sell one widget at a high price (high margin!), I should be unwilling to sell any at a lower price (but still high margin!) than the first one would sell for, because my rate of return on investment would go down (assuming the cost of producing a widget is dominated by variable costs). This does not describe the behavior of anything.

> This can't possibly be right. Why would you forego making extra money to keep your statistical average up? What are you getting from the average number?

You're serving the best interests of your shareholders, who have other things to be doing with their capital. They'd often rather invest $10 and get 10% than invest $20 and get 7%, even though that means getting $.10 rather than $.14.

You're comparing quarterly numbers with annual numbers.

Bitten by a denominator once again!