Will Wilkinson seems much more impressed by this Matt Welch column than I am (there were plenty of angles to criticize in Obama's prime time speech, but the fact that political rhetoric includes both predictive and conditional clauses? Really?). In particular, it's worth pushing back on one point. Welch snarks that "the president’s stimulus package 'will save or create 3.5 million jobs.' One of those, anyway!" Well, yeah. Job preservation is a much more powerful tool than job creation. On this, the economic literature is clear: A terminated worker's next job generally offers lower pay, lower benefit levels, and lower status. He's also less productive for the company: He doesn't always know the sector as well and there's a learning curve at any new place of employment. Meanwhile, keeping a company solvent through a hard few years is rather less expensive than inducing enough demand to create a whole new set of companies. Critics of this "create or save" formulation have a point in that it's a bit vague on the mix of creating and saving that they expect to be done. Fair enough. But if what you're interested in is the final unemployment rate -- and that is what we're interested in -- then it actually matters little whether you're creating or saving. The impact of either is that one more person is counted as employed rather than unemployed. But from the perspective of the taxpayer, the worker, and the company, saving existing jobs is better, cheaper, and more efficient than creating them. The more jobs the stimulus saves, the more effective it will have been. If we end up mainly having to create new jobs, that will mean our policy came too late.