Abstract

This paper documents the suite of models used by Norges Bank to estimate the output gap. The models are estimated using data on GDP, unemployment, inflation, wages, investment, house prices and credit. We evaluate the estimated output gap series in terms of its forecasting properties, its reliability and its cyclical sensitivity to various measures of demand and supply shocks. A simple un-weighted average of the models features a better forecasting performance than each individual model. In addition, it helps predicting inflation in pseudo real-time and exhibits limited variations when new data become available. The summary measure of potential output responds strongly and rapidly to permanent shocks and to narrative measures of technology shocks but, although to a more limited extent, also to transitory shocks.