Machine Learning Explainability in Finance: An Application to Default Risk Analysis

44 Pages Posted: 10 Aug 2019 Last revised: 12 Aug 2019

See all articles by Philippe Bracke

Philippe Bracke

Bank of England

Anupam Datta

Carnegie Mellon University - School of Computer Science

Carsten Jung

Bank of England

Shayak Sen

Carnegie Mellon University

Date Written: August 9, 2019

Abstract

We propose a framework for addressing the ‘black box’ problem present in some Machine Learning (ML) applications. We implement our approach by using the Quantitative Input Influence (QII) method of Datta et al (2016) in a real‑world example: a ML model to predict mortgage defaults. This method investigates the inputs and outputs of the model, but not its inner workings. It measures feature influences by intervening on inputs and estimating their Shapley values, representing the features’ average marginal contributions over all possible feature combinations. This method estimates key drivers of mortgage defaults such as the loan‑to‑value ratio and current interest rate, which are in line with the findings of the economics and finance literature. However, given the non‑linearity of ML model, explanations vary significantly for different groups of loans. We use clustering methods to arrive at groups of explanations for different areas of the input space. Finally, we conduct simulations on data that the model has not been trained or tested on. Our main contribution is to develop a systematic analytical framework that could be used for approaching explainability questions in real world financial applications. We conclude though that notable model uncertainties do remain which stakeholders ought to be aware of.

Keywords: machine learning, explainability, mortgage defaults

JEL Classification: C55, G21

Suggested Citation

Bracke, Philippe and Datta, Anupam and Jung, Carsten and Sen, Shayak, Machine Learning Explainability in Finance: An Application to Default Risk Analysis (August 9, 2019). Bank of England Working Paper No. 816, August 2019, Available at SSRN: https://ssrn.com/abstract=3435104 or http://dx.doi.org/10.2139/ssrn.3435104

Philippe Bracke

Bank of England ( email )

Anupam Datta

Carnegie Mellon University - School of Computer Science ( email )

5000 Forbes Avenue
Pittsburgh, PA 15213
United States

Carsten Jung (Contact Author)

Bank of England ( email )

Threadneedle Street
London, EC2R 8AH
United Kingdom

Shayak Sen

Carnegie Mellon University ( email )

Pittsburgh, PA 15213-3890
United States

Do you have negative results from your research you’d like to share?

Paper statistics

Downloads
2,520
Abstract Views
6,278
Rank
10,305
PlumX Metrics