John Cochrane writes in his blog, The Grumpy Economist, Christina Romer on Stimulus, 10/22/12
I don’t think anyone disagrees with the proposition that if the government takes money from residents of state A and splashes them on state B, the economy of state B improves. But this totally evades the whole issue: what about state A? This is the entirety of the stimulus debate: The government can transfer resources, but not get resources to fall from the sky. (Whether it taxes or borrows form the residents of state A, it’s still transferring resources.) Stimulus is supposed to raise aggregatedemand, not transfer demand from state A to state B. Yes, if the government builds a military base in the desert, GDP in that desert goes up. From this, stimulus raising the whole country does not follow.
The challenge of economics is to see the whole picture; the seen and unseen effects of a particular policy. This is the theme of the basic broken window fallacy of Bastiat. Technocrats act as if they can create wealth by merely transferring it. Stimulus by definition is temporary, and more investors are immune to short term incentives to stimulate production. A stimulus should deployed fast to be effective, but if such a such a short term expenditure is stimulative will it not be destimulative when it stops? The idea of a stimulus is to keep demand primed for a short period to overcome short term fears from market corrections. But over long time periods it is doomed and creates more problems than it solves. Stimulus is a seductive tool to those who think that wealth is actually generated by the government.