Reuters:  Which Tax Cuts Stimulate the Economy?, by David Cay Johnston:

Tax cuts are the key to job creation, or so Mitt Romney, running mate Paul Ryan and the 2012 Republican platform all say. But what does the empirical evidence show? Is the rhetoric in line with the known facts?

Studies examining the impact of cutting personal income tax rates on job growth or economic activity generally have been inconclusive, said Will McBride, chief economist for the Tax Foundation.

Owen M. Zidar, a graduate economics student at the University of California at Berkeley, and a former staff economist on the White House Council of Economic Advisers for President Obama, has taken another
crack at it, sifting through the data, using the National Bureau of Economic Research’s tax simulation model. Zidar looked at state level income and economic data. [Tax Cuts for Whom? Heterogeneous Macroeconomic Effects of Income & Payroll Tax Changes]

He reasoned that “if tax cuts for high income earners generate substantial economic activity, then states with a large share of high
income taxpayers should grow faster following a tax cut for high income
earners.” The data show that tax cuts at the top, though, do not not result in faster growth in states with more more high-earners.

“Almost all of the stimulative effect of tax cuts,” Zidar found, “results from tax cuts for the bottom 90%. A one percent of GDP tax cut
for the bottom 90% results in 2.7 percentage points of GDP growth over a two-year period. The corresponding estimate for the top 10% is 0.13 percentage points and is insignificant statistically.”

Here is the full abstract of the paper:

This paper investigates how tax changes for different income groups affect macroeconomic activity. Using historical tax return data from NBER’s TAXSIM, I construct a measure of who received (or who paid for) postwar tax changes for each income and payroll tax change that Romer & Romer (2010) classify as exogenous. At the national level, I aggregate tax changes for all taxpayers in the the bottom 90% and the top 10% of AGI and relate these aggregates to output, employment, and consumption growth. At the state level, I construct Bartik instruments for state tax shocks using national tax changes and each state’s share of high income taxpayers. If tax cuts for high income earners generate substantial economic activity, then states with a large share of high income taxpayers should grow faster following a tax cut for high income earners. I find that the negative relationship between tax changes and real GDP growth over a two year period is almost entirely driven by tax changes for the bottom 90%. The empirical relationship between tax cuts for the top 10% percent and job creation is negligible in magnitude, statistically insignificant, and much weaker than that of equivalently sized tax cuts for the bottom 90%.