For the feds right now, borrowing money is cheap. If the federal government can borrow at 0%, should that be considered the relevant cost of funds for making decisions on public investment projects? For example, suppose a proposed highway improvement would cost $10 million, is expected to last 10 years, and could be financed with a bond at 0% interest. Would we want the government to undertake the project, so long as it was expected to yield at least $1 million in benefits annually? Or should we insist, as a private firm probably would, on a bigger stream of benefits, to cover the opportunity cost of not investing that $10 million in something with a higher rate of return? This is a debate about the appropriate discount rate for making social decisions.
Tyler Cowen makes the latter opportunity cost argument, noting that if private investors make investment decisions based on a marginal private rate of return on capital of 10% or more, as they appear to, the government should too. Thus our highway project should yield annual benefits closer to, say $1.1 million.
Brad DeLong disagrees, using an argument from "first principles" on what the social discount rate should be. If we treat the utility of future generations equally with our own, which seems like the ethical thing to do, then the (risk-free) real discount rate should only be positive to the extent that people in the future will be better off than we are, and the utility of a dollar is worth less to a (future) person who has more of them. Typically these considerations imply a real social discount rate of about 2% or less.
This is an old debate. The Cowen position has been referred to as the descriptive approach, because it rests on actual market valuation by private agents in the economy. The DeLong position has been referred to as prescriptive, because it is, well, prescriptive. This debate has played an important role in economic analyses of optimal climate change policy. Because most of the costs of climate change mitigation occur now, and most benefits far in the future, the discount rate can make a huge difference in benefit-cost calculations. Indeed, using Cowen's preferred discount rate, the optimal policy response to climate change would be rather minimal.
Like DeLong, I prefer the prescriptive approach on ethical grounds, but Cowen's argument needs to be thought through. From a utilitarian point of view, private returns are social returns.* So we face the question of why private agents are underinvesting relative to the social optimum. DeLong offers some speculations as to why. But more to the point... If the government can borrow at less than 2%, and earn private returns of 10%, society would benefit if the feds borrowed a huge amount of money and invested heavily in the private economy. I.e., socialization. Risk is a factor, but the government is big and can do a lot of diversifying. So why not?
* In the absence of externalities or other market failures, of course.