General Information & Client Services
  • Americas: +1.212.553.1653
  • Asia: +852.3551.3077
  • China: +86.10.6319.6580
  • EMEA: +44.20.7772.5454
  • Japan: +81.3.5408.4100
Media Relations
  • New York: +1.212.553.0376
  • London: +44.20.7772.5456
  • Hong Kong: +852.3758.1350
  • Tokyo: +813.5408.4110
  • Sydney: +61.2.9270.8141
  • Mexico City: +001.888.779.5833
  • Buenos Aires: +0800.666.3506
  • São Paulo: +0800.891.2518

In this article, we compare the results of estimating retail portfolio risk parameters (e.g., PDs, EADs, LGDs) and simulating portfolio default losses using traditional – frequentist – methods versus Bayesian techniques. The statistical properties of the simulated risk parameter will have a significant effect on the shape of the portfolio loss distribution. Our results suggest that Bayesian estimations produce more robust estimators and result in risk parameters and loss distributions that are less volatile. Bayesian estimation has another key advantage: Posterior distributions for the model parameters can be leveraged to perform comprehensive portfolio loss simulation exercises taking into account model risk.

Default rate model set-up

We consider two common examples of retail portfolios, an auto loan book, and a credit card portfolio. Performance data is collected at a vintage/cohort level with quarterly frequency (i.e., quarterly cohorts of loans/accounts observed on a quarterly basis). The target variable to model is the vintage-level default rate, defined as the ratio of the number of accounts that have defaulted to outstanding accounts. Our methodology is in line with Licari and Suarez-Lledo (2013).1 Our target variable, (logit of) default rate, gets decomposed into three dimensions:

  1. Lifecycle component (seasoning of the accounts)
  2. Vintage quality (rank-ordering of the cohorts)
  3. Sensitivity to macroeconomic drivers

Parameters in (ii) and (iii) are assumed to be stochastic in the frequentist and Bayesian settings, while parameters (i) are assumed to be deterministic in both and fixed to the values found after performing the frequentist regression.

Model estimation results

Both frequentist and Bayesian methods produce similar average values for the key parameters. For auto loans, the results from both approaches are very similar. For credit cards, the Bayesian estimation has a significantly lower standard deviation in the parameter estimation than the frequentist standard errors, resulting in parameters that are less volatile and more precise and thus presenting a lower model risk when used in portfolio loss estimations.

Bayesian methods have the added advantage of producing posterior distributions for all parameters. The figures below illustrate the statistical shape of the posterior distribution for macroeconomic drivers and how these compare with frequentist parameters and 95% confidence intervals.

Model simulation results

The estimated models for the risk parameters are then used for the estimation of the portfolio loss distribution through a dynamic Monte Carlo simulation. Three distinct steps are considered:

  1. Macroeconomic scenarios are built using a Dynamic Stochastic General Equilibrium Model to produce forward-looking “states of the world” with quarterly updates (Licari and Ordonez-Sanz (2015)2).
  2. The Bayesian and frequentist models are then used to estimate risk parameters in each of these macroeconomic scenarios (e.g., conditional default rates).
  3. The portfolio is simulated dynamically (multiperiod default simulation) to estimate the cumulative loss distribution over nine future quarters.

For this last step, two different comparisons between the frequentist and Bayesian approaches were performed with and without model risk. The table below highlights the set-up of these two exercises, as well as key similarities and differences.

Figure 2 illustrates the set of macroeconomic scenarios used in the first step of the process for two different factors: unemployment rate and home price dynamics.

The estimated forward-looking default rates estimated in the second step of the process are shown below for both the frequentist and Bayesian approaches for the set of macroeconomic scenarios.

The first simulation exercise (macroeconomic risk only) shows consistency across both estimation methods. The simulated default rate distributions are fairly similar for both portfolios. They generate a tailed, asymmetric density with higher values at the block of stressed scenarios. The Bayesian method for credit cards seems to produce slightly less volatile default rate projections, but the overall shape of both densities is quite similar. The key difference appears when we move from exercise 1 to 2, adding model risk dimensions such as parameter volatility and error/residual properties.

Figure 4.2 drives home the key message of this section. Once we add the uncertainty coming from model risk to the simulation mechanism, frequentist and Bayesian methods produce very different outcomes. The higher precision of the Bayesian estimators flow into more concentrated, less volatile simulated default rates while still presenting the “fat-tails” that would be expected from the impact of very severe macroeconomic scenarios.

Effects on credit portfolio losses

The statistical properties of conditional default rates (conditional on a given macroeconomic scenario) influence the shape of the portfolio loss distribution. To quantify this effect, in step 3, dynamic Monte Carlo simulations are performed on both sets of conditional default rate distributions estimated in exercises 1 and 2. The charts below highlight the significant effect that these estimation results can have on the shape of portfolio losses. The severity of CCAR Adverse, Severely Adverse, ECCA S3 and S4 scenario as well as the VaR loss at 99.9% confidence levels are also shown for comparison purposes. In summary, Bayesian methods prove to be more stable, particularly after including model risk in the loss simulations.

Table 1. Macroeconomic parameters – Bayesian vs. frequentist estimations
Table 1A. Auto loan portfolio
Table 1B. Credit card portfolio
Macroeconomic parameters – Bayesian vs. frequentist estimations
Source: Moody's Analytics
Table 2. Alternative simulation exercises – macro only vs. fully-fledged
Alternative simulation exercises
Source: Moody's Analytics
Figure 1. Auto loan portfolio – posterior Bayesian distributions for macroeconomic parameters
Density functions (top) and box-plots (bottom). Red dots for frequentist betas, red vertical lines for frequentist 95% confidence intervals
Auto loan portfolio – posterior Bayesian distributions for macroeconomic parameters
Source: Moody's Analytics
Figure 2. Macroeconomic simulations – nine out-of-sample quarters
Macroeconomic simulations – nine out-of-sample quarters
Source: Moody's Analytics
Figure 3. Simulation exercise 1 – macroeconomic risk only – frequentist vs. Bayesian
Figure 3.1. Auto loan portfolio – across blocks of scenarios, five quarters out-of-sample (+Q5)
Simulation exercise 1 – macroeconomic risk only – frequentist vs. Bayesian
Figure 3.2. Credit card portfolio – across blocks of scenarios, five quarters out-of-sample (+Q5)
Credit card portfolio – across blocks of scenarios
Source: Moody's Analytics
Figure 4. Simulation exercise 2 – macroeconomic and model risks – frequentist vs. Bayesian
Figure 4.1. Auto loan portfolio – selected vintages, five quarters out-of-sample (+Q5)
macroeconomic and model risks
Figure 4.2. Credit card portfolio – selected vintages, five quarters out-of-sample (+Q5)
Credit card portfolio – selected vintages
Source: Moody's Analytics
Figure 5. Cumulative portfolio default losses – simulation exercise 1 – macroeconomic risk only – frequentist vs. Bayesian
Figure 5.1. Auto loan portfolio – after nine quarters (+Q9)
Cumulative portfolio default losses
Figure 5.2. Credit card portfolio – after nine quarters (+Q9)
Credit card portfolio – after nine quarters
Source: Moody's Analytics
Figure 6. Cumulative portfolio default losses – simulation exercise 2 – macroeconomic and model risks – frequentist vs. Bayesian
Figure 6.1. Auto loan portfolio – after nine quarters (+Q9)
Cumulative portfolio default losses
Figure 6.2. Credit card portfolio – after nine quarters (+Q9)
Credit card portfolio
Source: Moody's Analytics
Sources

1 See Licari & Suarez-Lledo, Stress Testing of Retail Credit Portfolios, Risk Perspectives Magazine, September 2013, Moody’s Analytics.

2 See Licari & Ordonez-Sanz, Multi-Period Stochastic Scenario Generation, Risk Perspectives Magazine, June 2015, Moody’s Analytics.

SUBJECT MATTER EXPERTS
As Published In:
Related Insights

U.K. Residential Mortgages Risk Weights: PRA Consultation Paper CP29/16

This paper presents best practices for addressing PRA Consultation Paper CP29/16.

October 2016 Pdf Dr. Juan M. Licari, Dimitrios Papanastasiou, Maria Valle del Olmo

Probability-Weighted Outcomes Under IFRS 9: A Macroeconomic Approach

In this article, we discuss development of a framework that addresses the forward-looking and probability-weighted aspects of IFRS 9 impairment calculation using macroeconomic forecasts. In it, we address questions around the practical use of alternative scenarios and their probabilities.

June 2016 WebPage Barnaby Black, Glenn LevineDr. Juan M. Licari

Complying with IFRS 9 Impairment Calculations for Retail Portfolios

This article discusses how to address the specific challenges that IFRS 9 poses for retail portfolios, including incorporating forward-looking information into impairment models, recognizing significant increases in credit risks, and determining the length of an instrument's lifetime.

June 2016 WebPage Barnaby Black, Dr. Shirish ChinchalkarDr. Juan M. Licari

Market Risk Stress Testing Models

In this presentation we present a two-stage process that generates consistent, transparent scenario-specific forecasts for all relevant market and credit risk instruments, ensuring cross-consistency between projections for macroeconomic and financial series.

December 2015 WebPage Dr. Juan M. Licari

Multi-Period Credit Risk Analysis: A Macro-Scenario Approach Presentation Slides

In this presentation, Dr. Juan Licari of Moody's Analytics will present an innovative framework for stochastic scenario generation that allows risk managers and economists to build multi-period environments, integrating conditional credit and market risk modeling to meet dynamic stress testing needs.

December 2015 Pdf Dr. Juan M. Licari

Market Risk Stress Testing Models Presentation Slides

In this presentation, Dr. Juan Licari presents a two-stage process that generates consistent, transparent scenario-specific forecasts for all relevant market and credit risk instruments, ensuring cross-consistency between projections for macroeconomic and financial series.

December 2015 Pdf Dr. Juan M. Licari

Multi-Period Credit Risk Analysis: A Macro-Scenario Approach

In this presentation, we present an innovative framework for stochastic scenario generation that allows risk managers and economists to build multi-period environments, integrating conditional credit and market risk modeling to meet dynamic stress testing needs.

November 2015 WebPage Dr. Juan M. Licari

IFRS 9 Impairment Webinar Series – Models for Implementation

This webinar discusses determining the best approaches for model development and governance for IFRS 9 Impairment calculations.

September 2015 WebPage Manuele Iorio, Dr. Juan M. Licari

Multi-Period Stochastic Scenario Generation

This article describes how to build consistent projections for standard credit risk metrics and mark-to-market parameters simultaneously within a single, unified environment: stochastic dynamic macro models.

June 2015 Pdf Dr. Gustavo Ordóñez-Sanz

Multi-Period Stochastic Scenario Generation

Robust models are currently being developed worldwide to meet the demands of dynamic stress testing. This article describes how to build consistent projections for standard credit risk metrics and mark-to-market parameters simultaneously within a single, unified environment.

May 2015 WebPage Dr. Juan M. Licari, Dr. Gustavo Ordóñez-Sanz

Model Governance

With regulators questioning the appropriateness of models, implementing a robust model governance is of paramount importance for banks. This article delves into governance best practices – including model definitions, inventory, categorization, and risk teams – and regulatory expectations.

November 2014 WebPage Dr. Gustavo Ordóñez-Sanz

Integrating Macroeconomic Scenarios into a Stress Testing Framework

This article describes the three principles that need to be understood and analyzed for banks to have a realistic chance of integrating alternative scenario work into their stress testing workflow.

November 2014 WebPage Dr. Juan M. Licari

Arbitrage-Free Scenarios for Solvency II

This article discusses a macroeconomic forecasting model that is able to generate arbitrage-free scenarios.

May 2014 WebPage Dr. Juan M. Licari, Dr. José Suárez-Lledó

Handling low default portfolios under stress

Regulators are challenging how to perform stress testing on low default portfolios by reviewing bank's PD models for RWA stress testing, in the absence of data they need to be convinced of the methodology used. In this Moody's Analytics webinar we put forward a statistical approach to stress testing low default portfolios with practical case studies

February 2014 Pdf Manuele Iorio, Dr. Juan M. Licari

Gauging the Risk of Europe's Banks: What Might the ECB Find?

The European Central (ECB) has begun a year-long comprehensive assessment of the Euro area banking system. In this webinar, Moody's Analytics seeks to provide a default data-driven context for the ECB's exercise and a preview for what is to come.

November 2013 WebPage Danielle Ferry, Dr. Juan M. Licari

Modelling and Stressing the Interest Rates Swap Curve

We present a two-step modelling and stress testing framework for the term structure of interest rates swaps that generates sensible forecasts and stressed scenarios out of sample. Our methodology is able to replicate two important features of the data: the dynamics of the spread across maturities and the alignment of the key swap rates tenor points to their corresponding government yields. Modern models of the term structure of interest rates typically fail to reproduce these and are not designed for stress testing purposes. We present results for the euro, the U.S. dollar, and British pound swap curves.

September 2013 Pdf Dr. Juan M. LicariOlga Loiseau-Aslanidi, Dr. José Suárez-Lledó

A Macroeconomic View of Stress Testing

This article discusses how developing deterministic scenarios form a macroeconomic view on stress testing that helps to uncover system or enterprise-wide vulnerabilities and assist banks in making more informed business decisions.

September 2013 WebPage Dr. Juan M. Licari, Dr. José Suárez-Lledó

Stress Testing of Retail Credit Portfolios

In this article, we divide the stress testing process for retail portfolios into four steps, highlighting key activities and providing details about how to implement each step.

September 2013 WebPage Dr. Juan M. Licari, Dr. José Suárez-Lledó

Reverse Stress Testing from a Macroeconomic Viewpoint: Quantitative Challenges & Solutions for its Practical Implementation

This whitepaper examines the challenge of multiplicity in reverse stress testing, where the same outcome can be obtained with multiple combinations of risk factors and economic scenarios.

October 2012 Pdf Dr. Juan M. Licari, Dr. José Suárez-Lledó

A Macro-finance View on Stress Testing

For most financial practitioners, stress-testing is a “must-do” activity, even if it is not a regulatory requirement. Such stress-testing encompasses a wide range of sophisticated and quantitative exercises, including assessments of market, credit and liquidity risks. This article discusses several approaches and outlines a foundation for a robust and consistent stress-testing framework.

May 2012 Pdf Andrea Appeddu, Dr. Juan M. Licari, Dr. José Suárez-Lledó

Reverse Stress Testing: Challenges and Benefits

Reverse stress testing is becoming recognised throughout the world for its benefits. This presentation explains what reverse stress testing is and what it can achieve, along with the challenges it presents. Here we show you why reverse stress testing can lead to a deeper understanding of an organisation's susceptibility to risk and why it is a valuable tool for any risk management strategy.

November 2010 Pdf Dr. Christian Thun, Dr. Juan M. Licari, Mark Zandi

Modeling and Stressing the Interest Rates Swap Curve

This article presents a two-step modeling and stress testing framework for the term structure of interest rates swaps that generates sensible forecasts and stressed scenarios out of sample. The results are shown for the euro, the US dollar, and British pound swap curves.

WebPage Dr. Juan M. LicariOlga Loiseau-Aslanidi, Dr. José Suárez-Lledó