Blanchard-Perotti (2002) An Empirical Characterization of the Dynamic Effects of Changes in Government Spending and Taxes on Output

Summary

Blanchard and Perotti characterize the dynamic effects of government spending and tax shocks on U.S. postwar activity using a "mixed structural-VAR / event-study" approach. Their central methodological contribution is an identification scheme that uses institutional information about the tax and transfer systems to pin down the automatic response of taxes and spending to output, and — because discretionary fiscal decisions cannot respond to output within the quarter — thereby to infer the structural fiscal shocks without a Cholesky ordering. Estimated on a trivariate quarterly VAR in net taxes, government spending, and output, the results consistently show positive spending shocks raising output and positive tax shocks lowering it; strikingly, both tax increases and spending increases have a strong negative effect on private investment.

Key Claims

• Identification via institutional elasticities. The reduced-form fiscal residuals mix three things: the automatic response of taxes/spending to output, the systematic discretionary response to output, and the structural fiscal shock. Using tax-code and transfer-system elasticities (the elasticity of net taxes to output is about 2; government purchases have essentially no automatic response), Blanchard and Perotti compute the automatic component directly. The systematic discretionary response is assumed absent within the quarter — at quarterly frequency, policymakers cannot observe output and legislate/implement a fiscal response fast enough — so the remaining residual variation identifies the structural shock.
• No Cholesky ordering needed. Because the contemporaneous elasticity of net taxes to output is known (not estimated), the ordering ambiguity that plagues recursive SVAR identification is avoided; the two orderings of the fiscal variables give nearly identical results.
• Spending multiplier. A positive government-spending shock raises output persistently, with a multiplier around one; private consumption rises in response — inconsistent with the neoclassical/RBC prediction that government spending crowds out consumption, and more consistent with a Keynesian view.
• Tax effects. A positive net-tax shock lowers output, with effects building over time.
• The nonstandard investment result. Both tax increases and spending increases strongly depress private investment — spending crowds out investment even as it raises consumption and output.
• Event-study cross-check. Specific fiscal episodes are examined to corroborate the VAR-based structural shocks, and deterministic trends / seasonal fiscal patterns are handled explicitly.

Concepts Introduced or Extended

• Fiscal Policy Shocks
• Structural Identification
• Impulse Response Function

Entities Mentioned

• Olivier Blanchard
• Roberto Perotti

Quotes

"Identification is achieved by using institutional information about the tax and transfer systems to identify the automatic response of taxes and spending to activity, and, by implication, to infer fiscal shocks."

"One result has a distinctly nonstandard flavor: both increases in taxes and increases in government spending have a strong negative effect on investment spending."

My Take

This is the fiscal counterpart to the recursive monetary-policy SVARs — but its identification is cleaner in one important respect: instead of assuming a timing/exclusion restriction and hoping it is credible, it imports a known number (the net-tax elasticity, ~2) from the tax code, so the contemporaneous fiscal response to output is calibrated rather than estimated. That single move is why the "Blanchard-Perotti approach" became the default identification in the fiscal-multiplier literature, the SVAR alternative to Ramey-Shapiro's narrative military-buildup dates. The consumption result (spending raises consumption) is the empirical fact that launched a decade of Keynesian-vs-neoclassical multiplier debates, and the method's main vulnerability — fiscal foresight, that agents and the VAR see tax/spending changes coming before they hit, so the "shock" is partly anticipated (Ramey 2011) — is exactly the invertibility/informational concern that later motivated narrative and proxy-SVAR fiscal identification. It pairs naturally with Monetary Policy Shocks as the two canonical policy-shock SVAR applications.