Title: GDP trend-cycle decompositions using state-level data
Authors: Manuel Gonzalez-Astudillo - Board of Governors of the Federal Reserve System (United States) [presenting]
Abstract: The aim is to develop a method for trend-cycle decomposition of GDP exploiting the cross-sectional variation of state-level GDP and unemployment rate data. In the model, each state's (log of) GDP is the sum of state-specific trend and cycle components, whereas the state unemployment rate is the sum of a trend and a cycle that is a function of the cycle of GDP in a way resembling an Okun's law. The state-specific GDP trend is a linear combination of a trend common across states and an idiosyncratic trend. The same is true for the state-specific unemployment rate trend. In a similar way, each state's GDP cycle is a linear combination of a cycle common to all the states and an idiosyncratic innovation. We estimate the model with Bayesian methods using quarterly data of GDP and the unemployment rate from 2005:Q1 to 2015:Q3 for the 50 states and the District of Columbia. Results show that the U.S. output gap reached about -6\% during the Great Recession, as opposed to conventional estimates of about -7.5\% (CBO-implied output gap), suggesting that the output trend was affected more than what conventional estimates obtain. Results also show that the output gap as of 2016:Q1 is about -1.5\%, compared to the CBO estimate of -2\%.