Kash has motivated me to do a little Google searching on something called “starve the beast” theory. How Tax Cuts Feed the Beast by Daniel Shaviro is a neat title but does not get to the point raised by Daniel Lambo. Shaviro does note:
On the campaign trail, President Bush is busy telling voters that he wants to keep making government ”smaller and more efficient” and that his opponent, John Kerry, wants a return to the days of big government. Forget the profligate spending of the past four years, he seems to suggest; instead, think of the many tax cuts that have been pushed through and how in time, as the saying goes, they will ”starve the beast.” But in fact the Bush tax cuts will do nothing of the kind. Counterintuitive though it may seem, they will inevitably end up increasing the size of government. When we talk about ”big government,” we shouldn’t be talking simply about how many agencies are in Washington or how many employees they have. Rather, we should consider government’s effects on society as a whole. If the government were to hire a lot of employees to redistribute wealth, everyone would agree that government had grown. But if the government instead used regulation for the same purpose, it might have fewer bureaucrats but it would not be smaller–it would simply have outsourced its ”big government” activity by making private workers subject to new government authority. In the same way, tax cuts can be used in lieu of spending to redistribute wealth and shape the economy. And, intentionally or not, that is exactly what Mr. Bush is doing.
It seems Shaviro agrees with Kash. Richard Vedder is often credited with Starve the Beast theory. The evidence put forth by Vedder is simply that higher taxes (T) are correlated with higher government spending (G). But does that mean T causes G or does not mean G causes T? It would seem that George von Furstenberg might argue G causes T – at least over the long-run given that the present value of taxes must equal the present value of government spending. This CBO discussion provides data, graphs, and a very good discussion of why Vedder’s correlations are not necessarily clear evidence of causation.