An Empirical Characterization of the Dynamic Effects of Changes in Government Spending and Taxes on Output

This paper by Olivier Blanchard and Roberto Perotti (QJE 2002; circulated from 1998) characterizes the dynamic effects of U.S. government spending and net tax shocks on output using a structural VAR identified not by the usual recursive timing assumptions but by institutional knowledge of the tax and transfer system: using quarterly data, the authors construct the automatic (non-discretionary) elasticity of taxes and spending to output from tax-code and transfer-program rules, and treat whatever is left over in fiscal innovations as policy shocks, arguing that legislative and implementation lags make it implausible that discretionary policy reacts to output within a single quarter. They find that positive government spending shocks raise output and positive net tax shocks lower it, with multipliers in most specifications “small, often close to one” and with no consistent evidence that the spending multiplier exceeds the tax multiplier as traditional Keynesian theory would predict. Decomposing the response by expenditure category, they find a result they call “distinctly non-Keynesian”: both spending increases and tax increases crowd out private investment, even though spending increases simultaneously crowd in private consumption – a combination the authors say is difficult to reconcile with either a standard Keynesian or a standard neoclassical model. The results are specific to postwar U.S. quarterly data (1960-1997 in the benchmark sample) and depend on the authors’ institutional elasticity estimates and on how the underlying trend is modeled (deterministic versus stochastic), though the paper’s robustness checks find the broad qualitative pattern stable across these choices.

Roberto Perotti | Macro Paper Warehouse

An Empirical Characterization of the Dynamic Effects of Changes in Government Spending and Taxes on Output