Principal Component Analysis for Cross-Sectionally Correlated Pricing Errors

Hyuksoo Kim, Saejoon Kim

Keywords: Unsupervised Learning, Optimization, Principal Component Analysis, Asset Pricing, Factor Pricing Model

Abstract: We propose a new estimator for factor pricing models which we refer to as Principal Component Analysis for Cross-Sectionally Correlated Pricing Errors (PCA-XC). Our estimator aims to find the factor pricing model that well explains the time-series variation of asset returns and well handles the correlations of cross-section of pricing errors that we present exist in real-world data. The proposed estimator is defined by a new regularized minimization problem in which finding a solution is difficult. This contrasts with other related estimators whose corresponding minimization problem admits an analytic solution. To this end, we propose an approximate algorithm that solves our proposed minimization problem based on the alternating least squares method.

Primary Area: unsupervised, self-supervised, semi-supervised, and supervised representation learning

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 8992