" It's hard to object to feel-good terms like "infrastructure" or "green" initiatives, but in a world of scarcity and choice, some spending is likely to be more beneficial, or at least less wasteful (Can you say Cash for Clunkers?), than another alternative. Do we choose bridges and overpasses instead of a high-speed rail network, restoring government buildings, more day-care centers and homes for senior citizens, or some elected official's pet project to placate his or her political base of support? Or perhaps shore up our human capital infrastructure—education, health—or address environmental concerns? In a world of finite resources, "Let's do them all" is simply not an option."
This does a good job describing the 'knowledge problem' plaguing stimulus & job creation. This is not an issue with market prices. Prices reflect tradeoffs based on the knowledge and specific preferences of millions of individuals, coordinate that information & provide incentives to act on it. (taking into account the multivariate array of alternatives, interactions, and consequences of each choice or action taken) Planned job creation & stimulus must rely on the relatively minute pool of knowledge at the disposal of a few voters, politicians, experts, or administrators, in the face of adverse incentive structures.
This doesn't mean to me that 'stimulus' spending has a zero multiplier. It means that it won't produce results that will allow us to get the most out of our scarce resources.