Title: On simple sum monetary statistics
Authors: Michael Ellington - University of Liverpool (United Kingdom) [presenting]
Abstract: The aim is to investigate how defective simple sum monetary aggregates can be. In doing so, we compare Divisia monetary statistics, which are measured using a micro-theoretic approach, with their simple sum counterparts. We provide a comprehensive reduced-form and structural analysis of time-varying parameter VAR models of the US economy using a range of monetary aggregates. Reduced-form results show a strong link between Divisia money and macroeconomic fundamentals at low and medium frequencies, whilst also providing substantial evidence supporting the conjecture that forecasters, and indeed central banks, should replace simple sum measures of money with their Divisia counterparts. Structural analysis uncovers economically significant and statistical differences in the structural impact coefficients of money within the monetary policy rule. Frequency domain variance decompositions report substantial contributions of monetary policy shocks to real GDP growth and inflation, both at an infinite horizon, and business cycle frequency. Although there are economic differences in the contribution these shocks conditional on the monetary aggregate used, we find no statistical differences. For policymakers, these results can help guide monetary policy reactions to recessions with an idea of the implications these shocks have on macroeconomic variation at different frequencies.