The lack of modesty among economists is pretty amazing. A couple of years ago, most economists knew that the Fed was great at handling interest rates and only moderate recessions would follow. Then many people figured that financial innovation was actually making the world a less risky place. Now, people seem to be basically fine with throwing about a trillion dollars to kick-start the economy. Temporary stimulus? Many of the projected jobs are in government, and it seems hard to imagine anyone firing them in a couple of years once the economy recovers.
With so many people rushing so quickly to adopt ideas that were considered academic nonsense not too long ago, it's worth being a little skeptical about the feasibility about a stimulus plan. Mankiw has a number of items up along this line of reasoning, which derive mainly from misreading papers by Obama appointee Christina Romer.
There's another line of criticism, however, aired by Posner. A stimulus will work in a Keynesian sense if the government uses up slack resources made available by the economic downturn, rather than using up productive resources at work elsewhere in the economy. By contrast, well-designed government investments are targeted at the sectors of the economy with the greatest potential. There's a problem here; it may be most expansionary to hire a construction worker in Las Vegas, but most productive to give a health-care worker there a different job.
The problem happens when you consider the ways in which people in a modern economy are highly specialized into specific sectors and tasks. This is related to the degree to which the economy runs on skilled workers and technology (even if it were as easy to switch jobs, the nature of skill endowments would make people want to stay within a particular area). It's unrealistic to simply throw together an infrastructure project and expect anyone to show up; sure, you might need a number of unskilled laborers, but the laid-off auto and finance workers are going to be unlikely to show up. People also don't particularly like to move to get a job; particularly if they have a house, family, or working spouse.
You can see where this might be a problem when you look at Obama's job creation plan. First of all, any sort of assumptions and projections you make are going to be iffy, especially when you're crafting an unprecedented policy proposal in uncharted waters. But they have to give Congress something, so I'll let that pass.
More troubling is where you see jobs being 'created.' State relief and business tax credit for new workers both make a lot of sense (though I hear the Democrats want to axe the tax credit). But are we really going to see hundreds of thousands of new jobs in education and health care appear ot of nowhere? Of course the long-term trajectory of these jobs is rising. But exactly for that reason, there aren't a lot of unemployed teachers. You really need to convert the newly-unemployed finance guys into teachers (and energy producers and infrastructure managers) if you want to get any sort of stimulus. But because of workforce specialization, it takes a long time to make that happen. Workers are no longer interchangable parts, but are increasingly sector and even firm specific. One way of looking at it is that many of the jobs on the stimulus agenda are female-specific, while the present crisis has affected male workers to a much greater degree.