Template-Type: ReDIF-Paper 1.0
Series: Tinbergen Institute Discussion Papers
Creation-Date: 2023-06-29
Number: 23-038/III
Author-Name: Jamie Cross
Author-Workplace-Name: University of Melbourne
Author-Name: Lennart Hoogerheide
Author-Workplace-Name: Vrije Universiteit Amsterdam
Author-Name: Paul Labonne
Author-Workplace-Name: Norwegian Business School
Author-Name: Herman K. van Dijk
Author-Workplace-Name: Erasmus University Rotterdam
Title: Bayesian Mode Inference for Discrete Distributions in Economics 
and Finance

Abstract: Detecting heterogeneity within a population is crucial in many economic and
financial applications. Econometrically, this requires a credible determination of
multimodality in a given data distribution. We propose a straightforward yet effective technique for mode inference in discrete data distributions which involves
fitting a mixture of novel shifted-Poisson distributions. The credibility and utility of our proposed approach is demonstrated through empirical investigations on
datasets pertaining to loan default risk and inflation expectations.
Classification-JEL: C11, C25, C81, C82, E00, D00
Keywords: Bayesian Inference, Mixture Models, Mode Inference, Multimodality, Shifted-Poisson.

File-URL: https://papers.tinbergen.nl/23038.pdf
File-Format: application/pdf
File-Size: 449.940 bytes
Handle: RePEc:tin:wpaper:20230038