Transcript *Do Tax Cuts Starve the Beast?* by Christina Romer and David

“Do Tax Cuts Starve the Beast?”
by Christina Romer and David Romer
Remarks by Steven J. Davis
University of Chicago
Brookings Panel on Economic Activity
April 2-3, 2009
Washington, DC
Starving the Beast –
What Is the Mechanism?
• Debt servicing requirements are a claim on
government revenues.
• Assumption: Rising marginal costs of raising
government revenue and non-increasing
marginal benefits of spending (for PM)
• Implication: OTE, a government that inherits
more debt spends less, net of debt servicing.
• Mechanism: Bequeath more debt to future
government, raise servicing costs, and thereby
limit future spending
Strategic Beast Starving
1. Current PM anticipates that future PM has a
stronger taste for public spending
–
–
–
Compared to situation where current PM expects to retain
power, current PM taxes less and runs a bigger deficit.
Why? To constrain future spending.
Persson and Svensson (1989, QJE)
2. Current PM anticipates that future PM wants a
different type of spending
–
–
–
Compared to situation where current PM expects to retain
power, current PM spends more on its preferred type of
spending and runs a bigger deficit.
Why? To constrain future spending on the type of
spending that current PM dislikes.
Alesina and Tabellini (1990, RESTUD)
No Starvation Diets in this Sample
• If deficit-financed tax cuts raise debt-GDP ratio by 5
percentage points and r=2% per year, then debt
servicing costs rise by (.05)(.02)=0.1% of annual
GDP
 Desired non-interest spending falls by 0.1% of GDP if MB
of G is flat, less if MB declines with G
• Even less strain on government revenue
requirements if tax cuts stimulate growth.
• More strain if higher debt causes r
• Conclusion: Mechanism is too weak for detection by
RR method in sample of postwar U.S. tax changes.
Another Mechanism
Tax changes are sticky – hard to reverse
•
•
•
Magnifies force of basic mechanism, because tax
changes implemented by current PM carry forward
Large # of tax changes documented by RR suggests
limited stickiness for “routine” tax changes
Stickiness could be a very important factor for outof-sample changes
– Elimination of corporate income tax system
– Adoption of VAT or national sales tax
What’s the Beast?
“Big G” or “Big Welfare State”?
•
•
•
Reagan opposed Big WS, favored defense
Many other advocates of beast starving are more
concerned about Big WS than Big Defense
If Big WS is target, a focus on total G can mislead
– Reagan cut taxes and initiated a persistent rise in
defense spending
– Both created fiscal pressures to restrain WS , but the
rise in defense spending muted decline of total G
•
Another issue: If defense spending is less
responsive to fiscal pressures, then beast-starving
effects load disproportionately onto WS spending.
– In line with Figure 5.d
Deficit-Fighting Tax Hikes
• Often accompanied by modest spending cuts
• Example: Clinton tax increase of 1993
• RR regard these changes as uninformative
about starve-the-beast hypothesis, because
they reflect a “switch to fiscal responsibility”
• But the need for deficit-fighting tax hikes and
spending cuts, when caused by deliberate tax
and spending decisions of previous PM, is
evidence for strategic beast-starving behavior.
– Persson and Svensson (1989)
Concerns about
Counterfactuals and Exogeneity
1. Suppose PM foresees rising political pressure for
public spending (e.g., because of population aging)
– In response, PM cuts taxes now to limit or slow future
spending growth.
– Spending grows, but less so because of tax cut
– RR interpret this pattern as evidence against starve-the-beast
effects, even if the tax cuts restrain spending growth.
2. Suppose tax cuts are more likely when PM foresees
strong growth and a lessening of fiscal pressures
–
–
In response, PM cuts taxes now and expands spending over
time.
RR interpret this pattern as evidence against starve-thebeast effects, even though both tax cuts and spending
growth reflect anticipations of strong growth
Summary
• An admirable effort, but not persuasive
–
–
–
–
Concerns about exogeneity and counterfactuals
Large standard errors
Too little variation to detect effects of mechanism
Issues related to defense spending vs. WS spending
• Figure 5.d says long run nondefense spending falls by
40% of tax cut amount (big standard errors)
• A priori reasoning implies the basic mechanism is
weak for tax change episodes in postwar U.S.
• Frequency of tax changes suggests that stickiness
mechanism also lacks force for routine changes
• Nothing here diminishes Becker-Mulligan-type
concerns about spending consequences of VAT
and other efficient taxes.
References
Becker and Mulligan, 2003, “Deadweight Costs
and the Size of Government,” J Law&Economics
Alesina and Tabellini, 1990, “A Positive Theory of
Fiscal Deficits and Government Debt,” RESTUD.
Persson and Svensson, 1989, “Why a Stubborn
Conservative Would Run a Deficit,” QJE.
Romer and Romer, 2009, “Do Tax Cuts Starve the
Beast?” BPEA, forthcoming.