Measuring the Natural Output Gap using Actual and Expected Output Data

An output gap measure is suggested based on the Beveridge-Nelson decomposition of output using a vector-autoregressive model that includes data on actual output and on expected output obtained from surveys. The paper explains the advantages of using survey data in business cycle analysis and the gap is provided economic meaning by relating it to the natural level of output defined in Dynamic Stochastic General Equilibrium models. The measure is applied to quarterly US data over the period 1970q1-2007q4 and the resultant gap estimates are shown to have sensible statistical properties and perform well in explaining inflation in estimates of New Keynesian Phillips curves.