The Behavioral Economics of the Stimulus

I knew that there was some serious behavioral economics behind the Obama team's plans to structure a tax rebate by reducing withheld tax! Jim Surowiecki has chapter and verse:

In the words of the behavioral economist Richard Thaler, people put different windfalls in different "mental accounts," which in turn influences what they do with the money...
People tend to base their spending not on their long-term earning potential or on their assets but on what they think of as their current income, an amount best defined by what's in their regular paycheck. When that number goes up, so does people's spending. In Thaler's words, "People tend to consume from income and leave perceived 'wealth' alone."
So what does this mean for making a rebate work? If you want people to spend the money, you don't want to give them one big check, because that makes it more likely that they'll think of it as an increase in their wealth and save it. Instead, you want to give them small amounts over time. And you want the rebate to show up as an increase in people's take-home pay, because an increase in steady income is more likely to translate into an increase in spending. What can accomplish both of these goals? Reducing people's withholding payments.

I do wonder whether and when behavioral economics will advance to the point at which it can actually predict specific different multipliers for different types of rebate. I'd love to see it make empirically-testable predictions -- not that economics generally is particularly good at those.