Research

We develop a new method for deriving high-frequency synthetic distributions of consumption, income, and wealth. Modern theories of macroeconomic dynamics identify the joint distribution of consumption, income, and wealth as a key determinant of aggregate dynamics. Our novel method allows us to study their distributional dynamics over time. The method can incorporate different microdata sources, regardless of their frequency and coverage of variables, to generate high-frequency synthetic distributional data. We extend existing methods by allowing for more flexible data inputs. The core of the method is to treat the distributional data as a time series of functions whose underlying factor structure follows a state-space model, which we estimate using Bayesian techniques. We show that the novel method provides the high-frequency distributional data needed to understand better the dynamics of consumption and its distribution over the business cycle.

Oil supply shocks shape U.S. inequality, and the type of oil supply shock matters for who bears the cost. In a Bayesian VAR that embeds the joint distribution of U.S. household income, consumption, and wealth alongside a standard oil-market block, sudden production shortfalls (Baumeister and Hamilton, 2019) raise inequality, whereas oil supply news (Känzig, 2021) mostly reduces it. These differences are not visible in the aggregate dynamics, which appear similar across shocks. A causal mediation analysis (Dufour and Wang, 2024) of these aggregate dynamics shows these shocks operate through distinct channels and it is this mechanism wedge, rather than the aggregate footprint, that determines who bears the cost. The analysis shares the view of Kilian and Lewis (2011) that policy responses should depend on the underlying causes of oil price shocks.

We provide a unified implementation of existing asymptotic theory that computes Conley-style spatial HAC standard errors for a wide range of commonly used (non-linear) estimators. We cover OLS, logit, probit, Poisson, and negative binomial regressions, as well as the fixed-effects estimators areg and reghdfe—extending commonly used publicly available routines (currently limited to linear models) to nonlinear M-estimators and fixed-effects workflows. We provide Stata and Python software implementing the procedure.