Matt Yglesias picks up on a discussion between Will Wilkinson and Joseph Heath about American conservatives' curious enthusiasm for providing "global public goods" (GPGs) in the form of enormous military spending to attempt to secure sea lines of communication (SLOCs) and do other things that are dubbed GPGs.
I think Matt is onto something bigger when he writes that
a considerable portion of American defense spending is genuinely wasteful. If we didn’t do it, it just wouldn’t be done. After all, it’s important to understand that excess capacity in military equipment is about as close as you can get to a real-world example of entirely wasteful public sector activity.
The economists tell us that one of the main properties of public goods is that they ought to be under-provided. As Matt writes, it seems like we're over-providing what are being called "public goods" here. To my mind, this strongly implies that they aren't public goods.
(Then again, if we're going to accept that the entire globe is the jurisdiction to which the U.S. government is supposed to be providing public goods, you're back to public goods -- that is, we're under-supplying the GPG of global security.)
While I'm not sure what my views are on traditional GPGs like a stable monetary or trading order, I'm very skeptical that anything related to security can be dubbed a GPG. The two key properties of public goods are nonrivalrousness and nonexcludability. Nonrivalrousness means that my consumption of the good doesn't conflict with yours. Nonexcludability refers to the idea that if you're living within the jurisdiction of the provider of the public good, it's impossible to opt out of consuming it.
Economists teach defense as the example of the quintessential public good. For two people living in a country, one's consumption of defense doesn't conflict with the other's and neither can be excluded from its benefits.
But I'm pretty sure you can't move from the idea of a bounded jurisdiction like that of a state to the entire globe and still have public goods in a meaningful sense. For example, the Japanese will recall from their experience in the 1930s that the control of SLOCs is very much excludable. Our inability to supply truly global security means that we have to pick and choose to whom we allocate resources. Our provision to one country of a formal alliance, for example, is very much rivalrous with neighboring countries' security.
More generally, in the context of the defense budget at home, I'm more inclined to think that a big chunk of U.S. defense spending constitutes a public bad: transfer payments from taxpayers to defense companies. Think about it for yourself; what is the marginal benefit to you of an extra F-22? An extra nuclear warhead? To whom is the social surplus (or more accurately, rents) allocated?
As is probably clear, my views aren't terribly well formed here, but the problem is an interesting one. Chris Preble discusses GPGs in The Power Problem, and a case against the defense budget as a pure public good is Jeffrey Rogers Hummel and Don Lavoie's "National Defense and the Public-Goods Problem," in Robert Higgs, ed., Arms, Politics, and the Economy (New York: Holmes & Meier, 1990), pp. 37-60.