Paper Image

Transfer learning panel data with auxiliary information

Published on:

29 August 2023

Primary Category:


Paper Authors:

Junting Duan,

Markus Pelger,

Ruoxuan Xiong


Key Details

Proposes target-PCA method to estimate latent factors for target panel using auxiliary data

Allows for missing data, weak factors, differences in dimensions and noise

Derives asymptotic distribution to select optimal weighting and impute values

Applies method to impute missing macroeconomic data at higher frequency

AI generated summary

Transfer learning panel data with auxiliary information

This paper develops a new method called target-PCA to estimate latent factor models for a target panel dataset. It leverages auxiliary panel data by combining it with the target panel in an optimal way. This allows dealing with missing data, weak factors, and differences in dimensionality and noise between panels. The method weights the target panel relative to auxiliary panels to ensure consistent estimation. It provides asymptotic theory to select weights that minimize variance and impute missing values precisely.

Answers from this paper


No comments yet, be the first to start the conversation...

Sign up to comment on this paper

Sign Up