Estimating Potential Output as a Latent Variable

Abstract

This article proposes a new method for estimating potential output in which potential real gross domestic product (GDP) is modeled as an unobserved stochastic trend, and deviations of GDP from potential affect inflation through an aggregate supply relationship. The output and inflation equations together form a bivariate unobserved-components model which is estimated via maximum likelihood through the use of the Kalman-filter algorithm. The procedure yields a measure of potential output and its standard error and an estimate of the quantitative response of inflation to real growth and the output gap.