Essays on economic policy analysis within the Bayesian econometrics framework

In this three-paper thesis, I study and develop econometric methodology to overcome problems in the application of the Bayesian inference principle in the empirical study of policy-relevant parameters. I focus on two main issues in Bayesian inference and estimation: (i) the identification of structural impulse responses under the local projections framework; (ii) the derivation of point and interval estimates of optimal taxes given a posterior distribution of structural parameters. Identification of structural impulse responses under local projections with external instrumental variables has been increasingly used to provide causal inference in empirical macroeconomics. I show the semi-parametric local projection has a full parametric form under an extended Wold representation, once econometricians "project out" the control variables. Using this form, I derive a Gibbs sampler to draw from the posterior of the impulse responses. I apply the proposed method to conduct an empirical analysis of two structural shocks in the U.S. economy: monetary policy shocks and marginal income tax shocks. In the first exercise, I conclude monetary policy shocks are non-neutral in a three years horizon. In the second exercise, I show the effect of permanent changes to tax income schedules on economic output is only temporary. The reason is that the economy responds to tax shocks by replacing the most taxed input, labor, with the least taxed one, capital. On the second issue, I show quantifying parameter uncertainty regarding optimal tax policy is non-trivial both from the perspective of econometricians and from the perspective of social planners. The conventional wisdom of ’plug-in’ the posterior of structural parameters provides biased point estimates and intervals with incorrect coverage. I propose a solution: econometricians need to incorporate the normative ii welfare functions when deriving point and interval estimates, while social planner needs to optimize expected welfare function, with expectation taken with respect to the posterior of unknown parameters. I provide a simulation study to support this approach.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Economics

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2023

Author dc:creator

• Morais Santos, Italo

Contributors dc:contributor

• Bernhadt, Dan M
• Amir-Ahmad, Pooyan
• Lee, Ji Hyung
• Howard, Greg
• Xie, Shihan

• Macroeconometrics
• Bayesian
• Quantitative Policy
• Monetary Policy
• Tax Policy
• Optimal Taxation

Statement dc:rights

• Copyright 2023 Italo Morais Santos

Language dc:language

en, eng

Handle dc:identifier

https://hdl.handle.net/2142/121478

OAI record oai:identifier

https://www.ideals.illinois.edu/oai-pmh?verb=GetRecord&metadataPrefix=etdms&identifier=oai%3Awww.ideals.illinois.edu%3A2142%2F121478