Debating Romer’s Research

Obama’s economic advisors have published estimates of the multipliers for fiscal stimulus, stating that $1 in new spending will raise GDP by $1.57, while $1 in tax cuts will raise GDP by only $0.99.

Greg Mankiw argues that economic research, including research done by Christina Romer (chair of Obama’s council of economic advisors), argues for a much lower multiplier for spending-based stimulus and for a much higher multiplier for tax cuts.

Nate Silver accuses Mankiw of deliberately misinterpreting Romer’s research, applying conclusions based on a study specifically on one type of tax cut to examine the effects of a different type of tax cut. Mankiw responds by arguing that Romer studied that type of tax cut because it’s easier to study, not because there’s reason to believe it behaves differently than other tax cuts, and Silver expands his argument in more detail (continuing to argue that Mankiw misinterprets Romer, while thankfully appearing to drop the accusation that the misinterpretation is deliberate).

Also responding to Mankiw’s original argument, David Henderson links to his own editorial examining the implications of Romer’s research, which he believes he has studied in more depth than Mankiw (and presumably Silver as well). Henderson’s read on Romer’s work is that it rejects not just tax cuts as a form of fiscal stimulus, but rather it rejects fiscal stimulus of all forms.

The money quote from Silver’s posts is:

So Romer and Romer identify two types of exogenous tax shocks, one associated with a larger-than-conventionally-assumed multiplier, and the other associated with a smaller-than-conventionally-assumed multiplier — perhaps even one in which the sign is reversed. In considering a third type of tax cut, an “endogenous” tax cut designed to stimulate growth during a recession, what basis does Mankiw to assume that it will behave more like the former than the latter?

Henderson’s article , although not responding specifically to Silver’s posts (indeed, Henderson wrote his article well before Silver wrote his), contains a response to this complaint.

One big problem with the Romers’ research, which they acknowledge, is that in their model one tax cut of a given magnitude is identical to another tax cut of the same magnitude. It doesn’t matter, in their model, whether the tax cut comes from a tax credit or from a cut in marginal tax rates. But, of course, it does matter.

The three types of tax changes Romer examine and to which Silver alludes are:

1. Tax cuts intended to support long-term economic growth. These cuts usually take the form of broad permanent reductions in tax rates, often with a focus on reducing taxes on savings and investment. It is these tax cuts that raise GDP by about $3 for every dollar of lost government revenue.
2. Tax increases intended to reduce deficits. These increases are usually a combination of raising tax rates and reducing or eliminating deductions and credits (closing loopholes). It is these tax increase which Romer finds have a lower-than-previously-thought effect on GDP.
3. Tax cuts to stimulate the economy in a recession and tax increases to prevent inflation during a boom. Tax increase to prevent inflation have only been put into effect once and are based on a now-long-discredited theory of what causes inflation. Tax cuts to stimulate the economy out of recession are much more common and usually take the form of short-term tax credits, often phased out for the upper-middle class and above.

Standard economic theory on the differing effects of types of tax changes, as Henderson notes, is that marginal tax rates are the most important factor in determining the economic drag of taxation. Type 1 tax cuts generally affect marginal rates the most, while Type 3 tax cuts generally affect marginal rates not at all, so Romer’s empirical findings appear to be perfectly consistant with standard economic theory.

The tax cuts Obama has been proposing are mainly expanded deductions and credits, while the tax cuts Mankiw has been proposing are mainly broad reductions in tax rates (an employer-side payroll tax holiday to encourage short-term hiring and a cut in taxes on savings and investment to support long-term recovery). Looked at in this light, the CEA estimate of a 0.99 multiplier for Obama’s tax cuts and Mankiw’s estimate of a 2.0-3.0 multiplier for his own proposed tax cuts are both correct applications of Romer’s research.