No, it is not zero, not even if government borrowing rates were literally at zero. Yet I’ve seen that claim a few dozen times in the last year or two, so let’s walk through some arguments that were fully standard by the 1970s.
Opportunity cost is ultimately defined in real resource terms, converted into value. So if the government borrows more money and mobilizes robots to do some work, that means fewer robots to do work elsewhere.
So what is the marginal private rate of return on capital? That’s a bottomless pit sort of question, but it’s not unusual to find sources which suggest a number for the average return in the range of 15% or so, see p..53 from this Stern School study here (pdf), with the median estimate running at about 12% (p.54). The papers from the 1980s found about the same, sometimes higher.
That’s way too high, says this stagnationist! Let’s instead cut it down to historic U.S. equity returns and say seven percent for the return at the margin.
Now, even today there are some unemployed resources. But most government fiscal policy works through well-known, fairly large contractors that at the margin have already well-established networks of capital and labor. So circa 2016, I don’t think that is a significant factor. Even in more down times, fiscal policy doesn’t always target unemployed resources so well.
That means a government project faces a seven percent hurdle rate. You may wish to up that for risk (the value of government output covaries positively with national income), and more yet for irreversibility. Let’s say that brings us up to a ten percent hurdle rate, and that’s being quite conservative. Sometimes irreversibility premia can multiply hurdle rates by 2x or 3x.
So that’s a (hypothetical) hurdle rate of ten percent, not zero percent. Of course it’s not unusual for private companies to use hurdle rates of twenty or more for their investment decisions.
There are further complications if the borrowing is financed by foreign finance capital. But there is still likely a real resource displacement in the home market as robots are shifted from one line of work to another. In addition, foreign ownership of the debt is about one-third of the total, noting the average and marginal here may diverge. Still, the domestic capital case seems to be the dominant effect.
You really can bicker about the right number here, and I’ve elided the question of whether some of the crowding out might come from consumption through a positive elasticity of robot supply; check the early papers of Martin Feldstein on related questions. But if someone tells you zero percent is the correct hurdle rate for government infrastructure investment, they are wrong.
Opportunity cost remains an underrated idea in economics.