Skip to main content
Mobile
  • Finance, Accounting & Economics
  • Global Business Management
  • Management, Leadership & Organisation
  • Marketing & Sales
  • Strategy
  • Technology & Operations
HS Talks HS Talks
Subjects  
Search
  • Notifications
    Notifications

    No current notifications.

  • User
    Welcome Guest
    You have Limited Access The Business & Management Collection
    Login
    Get Assistance
    Login
    Forgot your password?
    Login via your organisation
    Login via Organisation
    Get Assistance
Finance, Accounting & Economics
Global Business Management
Management, Leadership & Organisation
Marketing & Sales
Strategy
Technology & Operations
Practice paper

The validation of machine-learning models for the stress testing of credit risk

Michael Jacobs, Jr
Journal of Risk Management in Financial Institutions, 11 (3), 218-243 (2018)
https://doi.org/10.69554/SHKP5605

Abstract

Banking supervisors need to know the amount of capital resources required by an institution to support the risks taken. Traditional approaches, such as regulatory capital ratios, have proven inadequate, giving rise to stress-testing as a primary tool. The macroeconomic variables that supervisors provide to institutions for exercises such as the Comprehensive Capital Analysis and Review (CCAR) programme represent a critical input into this. A common approach to segment-level modelling is statistical regression, like vector autoregression (VAR), to exploit the dependency structure between macroeconomic drivers and modelling segments. However, linear models such as VAR are unable to model distributions that deviate from normality. This paper proposes a challenger approach in the machine-learning class of models, widely used in the academic literature, but not commonly employed in practice: the multivariate adaptive regression splines (MARS) model. The study empirically tests these models using Fed Y-9 filings and macroeconomic data, released by the regulators for CCAR purposes. The champion MARS model is validated through a rigorous comparison against the VAR model, and is found to exhibit greater accuracy and superior out-of-sample performance, according to various metrics across all modelling segments. The MARS model also produces more reasonable forecasts according to quality and conservatism.

Keywords: stress testing; CCAR; DFAST; credit risk; financial crisis; model risk; vector autoregression; multivariate adaptive regression splines; model validation

The full article is available to subscribers to the journal.

Already a subscriber? Login or review other options.

Author's Biography

Michael Jacobs, Jr is a senior vice-president at PNC Bank. He is a quantitative practitioner and researcher in the field of risk modelling and management. Prior to this, he was Principal Director at Accenture Consulting, Finance & Risk Advisory’s financial institutions, risk, regulatory and analytics practice. He has also worked as a senior economist with the US Office of the Comptroller of the Currency, a lead expert in enterprise and credit risk, and at J.P. Morgan Chase, where as a senior vice-president he led the empirical research and credit risk capital groups within the Risk Methodology Services division. He holds a doctorate in finance from the City University of New York, and is a Chartered Financial Analyst.

Citation

Jacobs, Jr, Michael (2018, August 1). The validation of machine-learning models for the stress testing of credit risk. In the Journal of Risk Management in Financial Institutions, Volume 11, Issue 3. https://doi.org/10.69554/SHKP5605.

Options

  • Download PDF
  • Share this page
    Share This Article
    Messaging
    • Outlook
    • Gmail
    • Yahoo!
    • WhatsApp
    Social
    • Facebook
    • X
    • LinkedIn
    • VKontakte
    Permalink
cover image, Journal of Risk Management in Financial Institutions
Journal of Risk Management in Financial Institutions
Volume 11 / Issue 3
© Henry Stewart
Publications LLP

The Business & Management Collection

  • ISSN: 2059-7177
  • Contact Us
  • Request Free Trial
  • Recommend to Your Librarian
  • Subscription Information
  • Match Content
  • Share This Collection
  • Embed Options
  • View Quick Start Guide
  • Accessibility

Categories

  • Finance, Accounting & Economics
  • Global Business Management
  • Management, Leadership & Organisation
  • Marketing & Sales
  • Strategy
  • Technology & Operations

Librarian Information

  • General Information
  • MARC Records
  • Discovery Services
  • Onsite & Offsite Access
  • Federated (Shibboleth) Access
  • Usage Statistics
  • Promotional Materials
  • Testimonials

About Us

  • About HSTalks
  • Editors
  • Contact Information
  • About the Journals

HSTalks Home

Follow Us On:

HS Talks
  • Site Requirements
  • Copyright & Permissions
  • Terms
  • Privacy
  • Sitemap
© Copyright Henry Stewart Talks Ltd

Personal Account Required

To use this function, you need to be signed in with a personal account.

If you already have a personal account, please login here.

Otherwise you may sign up now for a personal account.

HS Talks

Cookies and Privacy

We use cookies, and similar tools, to improve the way this site functions, to track browsing patterns and enable marketing. For more information read our cookie policy and privacy policy.

Cookie Settings

How Cookies Are Used

Cookies are of the following types:

  • Essential to make the site function.
  • Used to analyse and improve visitor experience.

For more information see our Cookie Policy.

Some types of cookies can be disabled by you but doing so may adversely affect functionality. Please see below:

(always on)

If you block these cookies or set alerts in your browser parts of the website will not work.

Cookies that provide enhanced functionality and personalisation. If not allowed functionality may be impaired.

Cookies that count and track visits and on website activity enabling us to organise the website to optimise the experience of users. They may be blocked without immediate adverse effect.