The Negative Mean Output Gap and the Symmetry Bias of Statistical Filters

Shekhar Aiyar,Simon Voigts
DOI: https://doi.org/10.1057/s41308-024-00257-8
2024-10-05
IMF Economic Review
Abstract:The output gap has a negative mean over time. More precisely, in a New Keynesian model with labor search frictions, downward nominal wage rigidity amplifies the decline in employment during recessions, causing a larger negative deviation from full employment output than the positive deviation during booms. But standard output gap estimation methods are constructed such that their estimates have a mean of zero. To analyze the resulting 'symmetry bias', standard filtering techniques are applied to simulated data from a NK model, allowing a comparison of the model's true gap with estimates. While filters are fairly accurate during periods of moderate macroeconomic fluctuations, they underestimate potential output substantially during deep recessions, when downward nominal wage rigidity is most binding. This is shown to lead to costly policy errors. If standard filter estimates inform countercyclical monetary or fiscal policy, symmetry bias weakens the policy response to recessions, resulting in higher output losses than under hypothetical knowledge of the true output gap. For example, focusing on the 25% deepest recessions, symmetry bias leads to inappropriately weak policy responses and premature tightening that amplifies output losses by more than a third on average.
economics,business, finance
What problem does this paper attempt to address?