“…Our paper also contributes to the literature that estimates statistical representations of individual earnings. 10 There are three ways prior studies have recovered income processes: (1) fitting statistical models of the earnings processes using GMM (e.g., Storesletten et al (2004), Karahan and Ozkan (2013), Guvenen et al (2015)), (2) fitting structural models of earnings processes using simulated method of moments (e.g., Blundell et al (2008), Guvenen and Smith (2014), and Madera ( 2016)), and (3) using Bayesian methods to estimate individual specific earnings processes (Geweke and Keane (2000), Jensen and Shore (2011), Nakata and Tonetti (2015), Gu and Koenker (2017), Borella, De Nardi, and Yang (2019) and Chatterjee, Morley, and Singh (2021)). 11 The first two methods do not allow researchers to recover realized shocks person-by-person, while the Bayesian methods, which encompass the filter presented in this paper, do.…”