Mountford-Uhlig (2009) What Are the Effects of Fiscal Policy Shocks?

Summary

This paper identifies the effects of fiscal policy using sign-restriction identification in a VAR, extending Uhlig's (2005) agnostic single-shock method to multiple simultaneous shocks. Mountford and Uhlig identify a government-revenue shock and a government-spending shock by imposing sign restrictions on the fiscal variables themselves while requiring both to be orthogonal to a generic business-cycle shock and a monetary-policy shock (each also sign-identified) — leaving the responses of GDP, consumption, investment, and real wages unrestricted (agnostic). They allow announcement effects (a shock that moves future but not current fiscal variables), and build the three canonical fiscal experiments — deficit-spending, deficit-financed tax cuts, and a balanced-budget spending expansion — as linear combinations of the two basic shocks. On U.S. quarterly data 1955–2000, deficit-financed tax cuts deliver the strongest GDP response, with a maximal present-value multiplier of about five dollars of total GDP per dollar of revenue cut, five years out. (First circulated as NBER WP 14551, 2008; published in the Journal of Applied Econometrics 24(6): 960–992.)

Key Claims

• Sign restrictions for multiple structural shocks. The identification generalizes Uhlig (2005): rather than pin down one shock, it jointly identifies a government-revenue shock and a government-spending shock by sign-restricting the fiscal variables' impulse responses, plus a business-cycle shock and a monetary-policy shock, all via sign restrictions.
• Orthogonality solves the automatic-response problem. Requiring the fiscal shocks to be orthogonal to the identified business-cycle and monetary-policy shocks filters out the automatic movements of taxes and spending in response to the cycle and to interest rates — one of the three core difficulties in fiscal identification.
• Fiscal shocks live in a two-dimensional space. "The" fiscal shock is not unique; macro fiscal policy is spanned by the revenue and spending shocks, and specific policies (balanced-budget expansion, deficit-financed tax cut, deficit spending) are different linear combinations of these two basic shocks — a clean way to define competing fiscal experiments.
• Announcement effects are modeled explicitly. Because fiscal policy is often announced before it is implemented, the method allows a current shock to change future fiscal variables without changing them today, so pre-implementation movements in macro variables are not misattributed.
• Agnostic on the variables of interest. No sign restrictions are placed on GDP, private consumption, private non-residential investment, or real wages — so their responses (and hence the multipliers) are outputs of the data, not assumptions.
• Main empirical finding — tax cuts dominate. Among the three scenarios on U.S. 1955–2000 data, deficit-financed tax cuts produce the best GDP outcome, with a maximal present-value multiplier ≈ 5 (dollars of cumulative GDP per dollar of the total revenue cut) at a five-year horizon. Government-spending multipliers are comparatively weak, and a deficit-spending expansion "crowds out" investment.

Concepts Introduced or Extended

• Fiscal Policy Shocks
• Sign Restriction Identification
• Structural Identification

Entities Mentioned

• Andrew Mountford
• Harald Uhlig

Quotes

"We use sign restrictions to identify a government revenue shock as well as a government spending shock, while controlling for a generic business cycle shock and a monetary policy shock."

"We find that deficit-financed tax cuts work best among these three scenarios to improve GDP, with a maximal present value multiplier of five dollars of total additional GDP per each dollar of the total cut in government revenue five years after the shock."

My Take

This is the sign-restriction answer to the fiscal-identification problem, and its methodological move — extending Uhlig's agnostic identification from one shock to a basis of shocks, with orthogonality doing the work that recursive ordering or institutional elasticities do elsewhere — is the lasting contribution. Defining fiscal experiments as linear combinations of a revenue and a spending shock is genuinely clarifying: it dissolves the "what is a fiscal shock?" ambiguity into a two-dimensional design space. The headline tax-cut multiplier of ~5 is large and became a reference point (and a lightning rod) in the multiplier debate; like all sign-restriction results it reports a set of admissible models, so the point estimates depend on the target/median convention and the responses stay wide. It sits naturally beside Blanchard-Perotti's institutional-elasticity identification as the agnostic alternative.