Traditional Keynesian economics says that during a recession, the government should run deficits to make up for the lack of demand in the economy. And Keynesian policy is built into the federal budget, which automatically increases spending on such things as unemployment insurance when the economy starts to tank. Unfortunately, as many have pointed out, the cumulative effect of the state budgets works in the other (pro-cyclical, as opposed to countercyclical) direction. States require their budgets to be balanced; revenues drop during a recession; and so states need to either raise taxes or cut spending or both. If California lays off a bunch of teachers, not only are they unemployed (a bad thing), but they stop buying stuff, and so stores and restaurants close, factories lay off shifts, and so on.
What's needed, as Matt Yglesias has argued, is to find a way to institutionalize the haphazard federal response to this problem. Christopher Edley steps up today with a proposal which I think is at least intriguing: let Treasury loan money to the states against future payments. He doesn't, however, specify a mechanism. Would states apply to the government for loans whenever they wanted? Would Treasury have any discretion about approving the loans? About setting repayment terms?
Andrew Samwick likes the idea, but cautions that "The federal government should be looking to get as much value for its spending, not just to spend money for the sake of spending it." Well, I'm not sure about that. I don't want to tread on the turf of economists, but it does seem to me that in a recession, just getting the money out there may be a higher priority than making sure it's used in the most efficient manner (and if that's what Samwick meant, then I agree).
As far as the politics of all this is concerned, however, it seems to me that there are two issues. One is to avoid introducing perverse incentives for the states; what governor wouldn't be tempted to borrow against future payments in order to inject money into the state economy in an election year, leaving the mess for later? And, the second issue would be the question of the political influence of the White House: if Treasury is given discretion, what's to keep an administration from helping friends and punishing foes? And if the answer is "nothing," then do we really want broad new influence for the White House over the state governments?
My feeling is that what's needed are automatic stabilizers, pegged to neutral criteria and activated with minimal discretion. Regular readers know that I tend in general to be all for political influence, but I do think that there's plenty of virtue in a little bit of autopilot here. The trick, I think, is to make it automatic on both sides: on the states asking, and on the Feds giving. After all, Congress would still have the power on top of this scheme to increase or decrease the money flowing to state governments, so politics would still be possible. The best way to think of it would be that the same (good) political interference would still be available, but the default setting would be rigged to "help the states" instead of "don't help the states."