Estimating the U.S. output gap with state‐level data

Abstract

Summary This paper develops a method to estimate the U.S. output gap by exploiting the cross‐sectional variation of state‐level output and unemployment rate data. The model assumes that there are common output and unemployment rate trend and cycle components, and that each state's output and unemployment rate are subject to idiosyncratic trend and cycle perturbations. I estimate the model with Bayesian methods using quarterly data from 2005:Q1 to 2018:Q2 for the 50 states and the District of Columbia. Results show that the U.S. output gap reached about negative 4.6% around the years of the Great Recession and was about 0.9% in 2018:Q2.

Estimation of short‐run predictive factor for US growth using state employment data

“…This useful precision in the state variable estimates due to the information from multiple indicators is consistent with the results of Basistha and Startz (2008) and Stock and Watson (2016). As González‐Astudillo (2019) noted the good precision of gap estimates during recessions, our confidence intervals of the employment factor are below zero during the recessions.…”

“…The states are Alabama, Georgia, Idaho, Illinois, Kentucky, Maine, Maryland, Massachusetts, Michigan, Minnesota, Montana, New Hampshire, New Jersey, Ohio, Rhode Island, and South Carolina. The diverse selection of these states is not particularly surprising as multiple studies by Owyang et al (2005), Hamilton and Owyang (2012), Guisinger et al (2018), Mumtaz et al (2018), Francis et al (2018), González‐Astudillo (2019), Mumtaz and Sunder‐Plassmann (2021) stress the presence of significant heterogeneity in economic structure and shock responses at the state level 4 . We then use all combinations of these 16 states with a minimum of at least two states to estimate the parameters of Equations (1) and (3)–(6).…”

Industrial Connectedness and Business Cycle Comovements

“…In many applications investigating cross-series cyclical correlation [e.g., Francis et al (2017) and González-Astudillo (2019)] the panels consist of countries or states and the models are estimated with data transformed into growth rates. Unlike the diversified economies in other studies, some of the industries in our sample may experience secular declines over part or all of the sample.…”

“…Examining the comovement of industries differs substantially from examining state or regional comovement [as in Hamilton & Owyang (2012) or González-Astudillo (2019)]. Because the latter are diversified economies in a common currency zone, they tend to have positive growth rates during expansions and negative growth rates during recessions.…”

Industrial Connectedness and Business Cycle Comovements

“…On the other hand, if clusters form across industries but within production networks, cycles may appear to be influenced by disruptions in the supply chain. 6 Examining the comovement of industries differs substantially than examining state or regional comovement [as in Hamilton and Owyang (2012) or González-Astudillo (2019)].…”