Former Treasury Secretary Lawrence Summers had interesting things to say, as usual, at a breakfast meeting this morning hosted by Gerald Seib and David Wessel of the Wall Street Journal.

In wide-ranging comments, he reiterated his view that premature fiscal contraction is costing the U.S. jobs and growth, and he argued for bringing forward infrastructure spending while interest rates make the real cost of borrowing roughly zero.

He was pressed on this by one questioner, who wondered whether the case for infrastructure spending was quite so clear-cut. Is it true that unemployment is especially high, as Summers had said, among the kind of workers who'd be hired for such projects? Unemployment is especially high in residential construction, to be sure, but that's a different industry: "It may be that the guys who hang drywall and string electrical cable are good candidates to run earth-movers and pour concrete, but that's not necessarily the case."

Summers agreed that the match wasn't perfect, but said it was close enough. From a labor market point of view, he said, now was as good a time to be making infrastructure investments as we would be likely to see in the next 10 or 20 years. Then he answered a question he hadn't really been asked, and drew a distinction between different kinds of infrastructure spending -- on one side, big new capital investments; on the other, maintenance.

He said accelerated maintenance spending probably has a higher rate of social return than investing in big new projects. Also, it can be geared up much faster, and the U.S. has a big backlog of this kind of spending. It's therefore far more suitable for stimulus purposes. But there's a problem. "You can't really name a filled-in pothole," he said. You can't "name an uncollapsed bridge that has been reinforced." It's "hard for a non-decaying school to be somebody's legacy because of a repair job. And so around the world, one finds within infrastructure investment a tendency toward overinvestment in the new and undermaintenance of the existing."

Until this problem is solved, supposing it ever can be, it seems to me to subtract a lot from the case for stimulus through infrastructure, desirable as this might be in principle. (Summers's comments reminded me of a 2009 paper by Bent Flyvbjerg, "Survival of the unfittest: why the worst infrastructure gets built -- and what we can do about it." Well worth reading.)

Summers was also interesting on corporate tax reform, especially as it affects international operations of U.S. companies. Current policy gives the U.S. the worst of both worlds, he said -- and he had a nice analogy to illustrate the point. If you have a library with a lot of overdue books out, you could announce an amnesty in the hope that some would be returned. Or you could promise there'd never be an amnesty, in the hope that some people would think they'd better fess up and return their books. The worst thing you could do is let people think an amnesty might be coming, but never actually have one. That way, you can be sure of getting no books back.

That's where taxation of foreign earnings now stands. "Clarity in either direction would magnify the incentive for the cash to come home," as Summers put it. He suggested an approach that would tax global income earned in the past but not yet repatriated at less than the existing corporate tax rate but more than zero rate implied by deferral.

Good idea.

(Clive Crook is a Bloomberg View columnist. Follow him on Twitter.)