Learning and Asset Acquisition Over the Lifecycle

Recent work by Kaplan and Violante (2014) has shown that the composition of household portfolios, between low return assets that can be freely transacted and higher return assets that have some measure of illiquidity, can affect the efficacy of fiscal policy. In this paper I consider a two asset lifecycle model with transaction costs and show that allowing individuals to learn about their idiosyncratic skill (and therefore their future income distribution) à la Guvenen (2007) over time yields starkly different asset allocation and overall savings decisions by households as compared to a more restricted income process typically considered in the literature. Using parameters estimated from the same underlying income data as the standard process, I show that a model with learning generates more liquid saving over the entire lifetime, suppresses illiquid saving early on, and increases illiquid saving later in the lifecycle. ∗I am grateful to Craig Burnside, Andrea Lanteri, and Kyle Jurado for their invaluable advice and continual support. I also thank Pietro Peretto, Cosmin Ilut, Matthias Kehrig, Francesco Bianchi, Nir Jaimovich, Curtis Taylor, Vasco Botelho, Soroush Ghazi, Linxi Chen, Marat Kussainov, and the participants of the Duke Macroeconomics Workshop and the Trade Dynamics Macroeconomcs Workshop Seminar series for their helpful comments and discussions.