The Wall Street Journal recently ran a piece by Bruce Bartlett arguing that a fiscal stimulus in the form of a tax rebate is no use, because people will save rather than spend the windfall.

It's an insult to Keynes even to call a tax rebate Keynesian economics. It should be called "feel good economics" because its only real effect is to make politicians feel good about themselves and buy re-election with the public purse.

People probably do expect too much from a fiscal stimulus under current circumstances, as I argued in an article earlier this week, but I think that Bartlett greatly overstates the case. I got an email from Larry Seidman of the University of Delaware, copying me a letter he sent to the WSJ responding to Bartlett. Since, so far as I know, it hasn't yet run, I am posting it here with Larry's permission. He makes points that advance the discussion and deserve to be better understood--especially concerning the importance of (a) cutting the payroll tax and (b) sustaining the treatment.

In "Feel-Good Economics" (Jan 19-20), Bruce Bartlett writes, "In short, there is virtually no empirical evidence that tax rebates are an effective response to economic slowdowns."  He may not be aware of the empirical study using data from the Consumer Expenditure Survey published in the leading peer-reviewed technical journal of the economics profession, the American Economic Review, in its December 2006 issue entitled "Household Expenditure and the Income Tax Rebates of 2001" co-authored by David Johnson (U.S. Census Bureau), Jonathan Parker (Princeton), and Nicholas Souleles (Wharton, U of Pennsylvania) who found that "households spent 20 to 40 percent of their rebates on nondurable goods during the three-month period in which their rebates arrived, and roughly two-thirds of their rebates cumulatively during this period and the subsequent three-month period.  The implied effects on consumption demand are substantial.  Consistent with liquidity constraints, responses are larger for households with low liquid wealth or low income."  The last sentence implies that it is crucial to rebate payroll tax as well as income tax in order to reach low-income households.

He also may not be aware of another empirical study published in the journal Business Economics in its July 2006 issue entitled "A Temporary Tax Rebate in a Recession: Is it Effective and Safe?" in which my co-author Kenneth Lewis and I, using a macroeonometric model by Ray Fair (Yale) and the empirical results from a 2003 AER article by Joel Slemrod (U of Michigan) and Matthew Shapiro (U of Michigan), estimated that if the 2001 rebate had been twice as large ($1,200 instead of $600) and had been repeated four times (every quarter for one year), it would have reduced the unemployment rate a full percentage point (from 6.0% to 5.0%).  This empirical estimate implies that in today's larger economy (GDP $14 trillion instead of $10 trillion), a $1,700 payroll-plus-income-tax rebate repeated four times (every three months) would reduce the unemployment rate from 6.0% to 5.0%, saving 1.5 million jobs (with 5.0% there are 7.7 million unemployed; with 6.0% there would be 9.2 million unemployed).  This would cost roughly $100 billion per quarter or $400 billion for the year--about 3% of GDP.

Thus, contrary to Bartlett's assertion, empirical evidence suggests that as long as the dosage of a payroll-plus-income-tax rebate is strong (twice as large as in 2001), sustained (repeated every three months until the economy has recovered), and stopped (once recovery is complete), it is effective and safe medicine for a slowing economy.

Incidentally, Larry reminded me that I wrote a piece for The Economist in 2002 about an earlier article of his on automatic fiscal stabilisers. If you are curious, you can read it here. It had slipped my mind. This is not the only issue that cycles up and down with the economy.