Starve-the-Beast: It’s Complicated

by Reihan Salam

Ramesh Ponnuru’s new Weekly Standard article on “starving the beast” suggests that it’s not curbing tax increases as such that constrains public expenditures, but rather curbing tax increases on middle-income households. In the wake of the tax deal that resolved the fiscal cliff mini-crisis, he sees new battle lines:

Liberals will continue to want higher taxes and may even try, should the tides of politics shift sufficiently in their favor, to raise them on the middle class. The deal only means that future tax-hikers will have to raise middle-class taxes from a lower level than they would have had we fallen off the cliff to stay, and they will have to fight harder for their gains. Those are not small things.

Ramesh acknowledges that recent years have cast doubt on “starve the beast” theory, reviving interest in “serve the check” theory, i.e., the belief that the electorate will only favor spending restraint when it is obligated to pay for spending in the form of increased taxes. Yet he suggests that low taxes on middle-income households serve a special role:

The track record shows that neither starving the beast nor serving the check reliably constrains government spending. But if you believe that at some point, not all that far off, a rebalancing of the federal budget is both necessary and inevitable, and all of the budget deals and fights of any given year are actually attempts to influence the shape of that eventual settlement, then matters are pretty simple: Lower taxes push in the direction of lower spending, and higher spending in the direction of higher taxes. And the lower taxes that matter most for the shape of that settlement are the taxes on the middle class.

To elaborate, middle-income households are “where the money is” — tax increases on high-income households can generate revenue, but these households tend to have more volatile incomes, and so a highly progressive tax base is likely to also be volatile. And one problem with raising marginal tax rates is that the disincentive effect of doing so tends to outweigh the resulting revenue gains, as workers are sensitive to marginal incentives even though their first dollars might be taxed at a much lower rate. This is one reason why a more effective strategy for raising revenue might be lowering the threshold at which the top rate kicks in rather than raising the top rate, though of course this would be more politically problematic. 

So the harder it is to raise taxes on middle-income households, the harder it is to generate the kind of revenue required to make existing health entitlements fiscally sustainable without significant structural reform. At some point, the revenue gains that flow from raising taxes on the highest earners will be outweighed the larger economic consequences. This is not to suggest that we are necessarily that close to the top of the Laffer Curve. But as Arpit Gupta observed in his NR article on “Tax Rates and Economic Growth,” this isn’t the only relevant question:

Until we reach the top of the Laffer curve, higher taxes bring in more revenue, and thus help to control our debt. But as we saw above, taxes also affect economic activity, and thus our GDP. So taxes affect the debt-to-GDP ratio in both the numerator and the denominator. Even if a tax increase raises revenue, it may lower GDP in a way that makes achieving long-run fiscal balance harder. 

Daniel Mitchell of the Cato Institute assesses the (meager) empirical evidence on the “starve the beast” thesis and he makes an important observation: “you can’t starve the beast if you don’t maintain the diet.” That is, as Christina Romer and David Romer note in a broad survey, “roughly three-quarters of a long-run tax cut is typically undone by legislated tax increases of various sorts within five years.” And so, according to Mitchell, we haven’t had a good test of the hypothesis.