Research Papers

We develop a model-based approach to constructing investment grade and high yield corporate bond portfolios that both outperform their respective prevalent benchmark indices and popular Exchange Traded Funds (ETFs) with better risk-return profiles, i.e., higher returns with lower or similar risk. More importantly, the outperformance is obtained after controlling for credit risk, duration risk, and downside risk. The outperformance is robust across a number of different specifications of the strategy and over time. We also achieve outperformance using relatively smaller, more realistic portfolios. These model portfolios can be potentially converted into new fixed income indices or ETFs. Our model-based approach utilizes Moody’s Analytics’ EDF™ (Expected Default Frequency) credit measures and Fair-value Spread (FVS) valuation framework as powerful tools to control for credit risk and to exploit relative value in the bond market.

Author: Zan Li, Jing Zhang, Christopher Crossen
Date: April 19, 2012

A major source of firm funding and liquidity, credit lines can pose significant credit risk to the underwriting banks. Using a unique dataset pooled from multiple U.S. financial institutions, we empirically study the credit line usage of middle market corporate borrowers. We examine to what extent borrowers draw down their credit lines and the characteristics of those firms with high usage. We study how line usage changes with banks’ internal ratings, collateral, and commitment size and through various economic cycles. We find that defaulted borrowers draw down more of their lines than non-defaulted borrowers. They also increase their usage when approaching default. Risky borrowers tend to utilize a higher percentage of their credit lines as well.

Author: Janet Yinqing Zhao, Douglas W. Dwyer, Jing Zhang
Date: December 12, 2011

The recovery amount from a defaulted financial instrument can be measured by either ultimate recovery (UR) or post-default value (PDV). While these two measures have the same expected value, UR variance is usually greater than PDV variance. In most cases, the appropriate measure of variance to use in an economic capital framework is somewhere between UR and PDV variance, but is much closer to PDV variance. This paper proposes two approaches to estimating recovery variance based on UR variance and PDV variance, respectively. We find that the impact of recovery variance parameterization on economic capital can be non-trivial, especially when PD-LGD correlation (PLC) is accounted for. In this paper we demonstrate that two common parameterizations can result in capital differences of over 15% in a realistic setting.

Authors: Amnon Levy, Qiang Meng, Douglas Dwyer and Irina Korablev
Date: November 17, 2011

The RiskCalc v3.2 Canada model provides a measure of default risk for Canadian private firms. We develop, calibrate, and validate the model using a large dataset of local financial statements and defaults. We released our original RiskCalc v3.1 Canada model in 2004 and RiskCalc v3.2 Canada in 2009. This paper presents the underlying research, model characteristics, data, and validation results for the v3.2 Canada model’s current performance. Our latest validation includes the data used in developing the original model, as well as newly received financial statement data through 2010. Improved data coverage enables us to refine our financial statement model and achieve a very robust prediction model of private firm default behavior. Our recent results show that the model effectively measures default risk, both in-sample and out-of-sample, across industry, size, and different time periods.

Author: Yu (Lucy) Jiang, Irina Korablev and Douglas W. Dwyer
Date: November 7, 2011

In this paper, we validate the performance of the Moody’s Analytics Public EDF™ (Expected Default Frequency) model for global financial firms during the last decade, including the recent credit crisis and its recovery period. We divide the decade into two sub-periods: an early period, 2001–2007, and a later one, 2008–2010, and then compare the model’s performance during these two periods. We focus on the model’s ability to prospectively differentiate between defaulters and non-defaulters, its comparison to agency credit ratings, the timeliness of its default prediction, and its accuracy of levels.

Author: Christopher Crossen and Sue Zhang
Date: October 2011

We validate the performance of the Moody’s Analytics Public EDF™ (Expected Default Frequency) model for Asian-Pacific and Japanese corporate firms during the last decade, including the recent credit crisis and its recovery period. We divide the decade into two sub-periods: an early period, 2001–2007, and a later one, 2008–2010, and then compare the model’s performance during these two periods. We focus on the model’s ability to prospectively differentiate between defaulters and non-defaulters, the timeliness of its default prediction, and its accuracy of levels.

Author: Christopher Crossen and Xu Zhang
Date: October 2011

We validate the performance of the Moody’s Analytics Public EDF™ (Expected Default Frequency) model for European corporate firms during the last decade, including the recent credit crisis and its recovery period. We divide the decade into two sub-periods: an early period, 2001–2007, and a later one, 2008–2010, and then compare the model’s performance during these two periods. We focus on the model’s ability to prospectively differentiate between defaulters and non-defaulters, the timeliness of its default prediction, and its accuracy of levels.

Author: Christopher Crossen and Sue Zhang
Date: October 2011

Analysts often find themselves working with less than perfect development and/or validation samples and data issues typically impact the interpretation of default prediction validation tests. Discriminatory power and calibration of default probabilities are two key aspects of validating default probability models. Both are susceptible to data issues. In this paper, we look at how data issues affect three important power tests: the Accuracy Ratio, the Kolmogorov-Smirnov test, and the Conditional Information Entropy Ratio, as well as how data issues affect the Hosmer-Lemeshow test, a default probability calibration test. We employ a simulation approach that allows us to assess the impact of data issues on model performance when the exact nature of the data issue is known.

Author: Heather Russell, Qing Kang Tang, Douglas Dwyer
Date: August 2011

Through-the-Cycle EDF (TTC EDF) credit measures are one-year probabilities of default that are largely free of the effect of the aggregate credit cycle, primarily reflecting a firm’s enduring, long-run credit risk trend. TTC EDF measures are useful in applications in which a stable PD input is desirable, and for which the expected cost of adjusting credit exposures as PD signals change outweighs the expected cost of negative credit events (such as default).

Author: David T. Hamilton, Zhao Sun andMin Ding 
Date: August 2011

In this paper, we validate the performance of the Moody’s Analytics Public EDF™ (Expected Default Frequency) model for North American corporate firms during the last decade, including the recent credit crisis and its recovery period. We divide the decade into two sub-periods: an early period, 2001–2007, and a later one, 2008–2010, and then compare the model’s performance during these two periods. We focus on the model’s ability to differentiate between good and bad firms, its comparison to agency credit ratings, the timeliness of its default prediction, and its accuracy of levels.

Author: Christopher Crossen, Shisheng Qu and Xu Zhang
Date: May 20, 2011

Commercial real estate (CRE) exposures represent a large share of credit portfolios for many banks, insurance companies, and asset managers. It is critical that these institutions properly measure and manage the credit risk of these portfolios. In this paper, we present the Moody’s Analytics framework for measuring commercial real estate loan credit risk, which is the model at the core of our Commercial Mortgage Metrics (CMM)™ product. We describe our modeling approaches for default probability, loss given default (LGD), Expected Loss (EL), and other related risk measures.

Author: Jun Chen and Jing Zhang
Date: May 3, 2011

In this presentation we examine the strengths of a risk calculation model that assesses localized accounting practices of individual countries within the wider context of the credit cycle. The model takes account of liquidity, profitability, activity, leverage, growth variables and other integrated factors to deliver objective results. Here we put the spotlight on exactly what this model can do and how it works.

Author: Douglas Dwyer
Date: May 2011

On any given day, credit analysts monitor multiple names. Some names will have been reviewed recently, but not all. Some names are easily traded out of, while some names are more difficult to hedge. Some names represent large exposures while others represent small. Some are known high credit risks while others are low credit risks. The risk profile of some exposures may have changed recently while others remained unchanged. How to triage? Which names should the analyst review first? Next? Ultimately, how does the institution value a credit review? This paper derives an optimal monitoring strategy, under the assumption that there is a fixed cost for reviewing a loan. The framework can be embedded within a dynamic competitive equilibrium in which prices reflect public information. We calibrate the framework using parameter values with empirical interpretations. In the dynamic setting, we show that in specific circumstances, a “mid-year” review can produce incremental value equal to 1.3% of the exposure size.

Author: Douglas W. Dwyer
Date: April 2011

Financial institutions, particularly banks, were at the heart of the credit crisis and subsequent  recession, and defaulted at unprecedented rates. It will be a long time before names like Lehman  Brothers, Bear Stearns, and Northern Rock fade from the memories of investors and risk managers. Not surprisingly, the experience has redoubled interest in finding effective and efficient ways to provide early warning of credit distress for such entities.

Authors: David W. Munves, CFA, Allerton (Tony) Smith, David T. Hamilton, PhD
Date: December 15, 2010

There are advantages to measuring credit risk quantitatively, when possible. Nevertheless, qualitative factors may add information, because some credit risk determinants cannot be captured by quantitative measures.

Authors: Douglas Dwyer, Heather Russell
Date: November 15, 2010

The end-to-end automation of commercial loan origination is certainly not a new idea, yet it remains an elusive goal for many financial services institutions. Most banks continue to operate using processes that are highly manual. They have, at best, point solutions to automate particular aspects of the loan origination life cycle but little in the way of true integration or standardization across commercial lending products.

Author: Susan Feinberg
Date: November 2010

Corporate credit is an important investment class with potentially attractive returns. The asymmetric nature of corporate bond returns implies that investing in credit is not about picking “winners,” but rather about avoiding “losers” (i.e., defaults and spreads blow-ups). Investors can utilize quantitative measures of credit risk to minimize the risk of such events.

Authors: Zan Li, Jing Zhang
Date: September 14, 2010

This paper presents a study of momentumimpact within the Moody's Analytics EDF™ (Expected Default Frequency) credit measure on patterns of default risk and rating changes. We define momentum as the significant rise or fall of an issuer's one-year EDF level during a one-year horizon. The main question motivating our study is whether or not such momentum can provide useful information about future credit events, thus improving the utility of the EDF credit metric for risk managers and investors.

Authors: Ozge Gokbayrak, Lee Chua
Date: July 2, 2010

In this paper, we present a framework that links two commonly used risk metrics: default probabilities and credit spreads. This framework provides credit default swap-implied (CDS-implied) EDF (Expected Default Frequency) credit measures that can be compared directly with equity-based EDF credit measures. The model also provides equity-based Fair-value CDS spreads (FVS) that can be compared directly with observed CDS spreads.

Authors: Douglas Dwyer, Zan Li, Shisheng Qu, Heather Russell and Jing Zhang
Date: March 11, 2010

In this paper we study the effectiveness of using equity returns for corporate default prediction. Specifically, we analyze whether using equity return information alone can yield similar performance to EDFs in default prediction. We find that the answer is no.

Author: Zhao Sun
Date: January 27, 2010

In this paper, we validate the performance of the Moody’s Analytics EDF™ (Expected Default Frequency) model during the recent credit crisis.

Authors: Ozge Gokbayrak, Lee Chua
Date: November 18, 2009

This document outlines the validation results for the RiskCalc v3.1 U.S. Banks model, and highlights the deteriorating financial ratios present in the banking sector. We contrast trends of key risk measures to those of the savings and loan crisis of the late 1980s and early 1990s. We also explore the speed and nature of recent bank failures and demonstrate the model?s strong performance in light of this rapidly changing environment.

Authors: Douglas W. Dwyer, Daniel Eggleton
Date: October 8, 2009

In this paper, we validate the Moody's KMV RiskCalc v3.1 United States private firm default model. We show that the EDF™ (Expected Default Frequency) produced by the model continues to rank order risk effectively by providing substantial discriminatory power across multiple cuts of the data.

Authors: Douglas W. Dwyer, Daniel Eggleton
Date: September 2, 2009

In this paper, we validate the performance of the Moody's KMV EDF™ (Expected Default Frequency) model during the recent credit crisis. We analyze the model performance during the past two years, and compare this performance to the model's longer history (1996-2006).

Authors: Irina Korablev, Shisheng Qu
Date: June 26, 2009

Moody’s KMV LossCalc is the Moody's KMV model for predicting loss given default (LGD). In April 2009, Moody’s KMV introduced its newest LossCalc model, LossCalc v3.0. Building on the foundation of its predecessors, this model provides users with a systematic approach to estimating recovery on a given issue. In addition, it accounts for how geography, industry, credit cycle stage, debt type, standing in the capital structure, collateral type, and the firm’s credit quality influence recovery. LossCalc v3.0 provides a term structureof recovery; the estimated LGD depends on when the default occurs. Specifically, the timevarying drivers of recovery have a smaller impact on predicted recovery the further into the future the default event occurs. Further, the model provides a method for stressing the LGD by reporting estimates of what the recovery on a specific loan would have been at different stages of the business cycle. While the model is based on post-default prices, we show the relationship between recovery measured using post-default prices, and recovery measured by ultimate recovery (i.e., the present discounted value of the cash flows from the loan if the loan is held through the resolution process).

Authors: Douglas Dwyer, Irina Korablev
Date: April 9, 2009

In this paper, we validate the performance of Moody’s KMV EDF™ credit measures in its timeliness of default prediction, ability to discriminate good firms from bad firms, and accuracy of levels in three regions: North America, Europe, and Asia.

Authors: Irina Korablev, Douglas Dwyer
Date: September 10, 2007

This paper proposes a theoretical framework to account for systematic risk in recovery and to address the correlation between the firm’s underlying asset process and recovery. Under the proposed framework, the expected value in default under the risk neutral measure can be expressed as a linear function of the expected value under the physical measure. This allows for a simple mapping between expected recovery observed in the data and a measure that can be applied when using risk neutral valuation methods.

Authors: Amnon Levy, Zhenya Hu
Date: April 20, 2007

In July 2002, the International Accounting Standards Board (IASB) published a new set of accounting policies to be adopted by all firms incorporated in the European Union as of January 1, 2005. The new rubric, called the International Financial Reporting Standard (IFRS), replaced a variety of preexisting standards in France, Germany, the United Kingdom, and other European countries, each of which had a slightly different set of Generally Accepted Accounting Principles (GAAP) standards.

Authors: Adam Rapp, Shisheng Qu
Date: April 18, 2007

Recent turmoil in the subprime mortgage market claimed several victims, notably New Century Financial Corporation (NEWC), which filed for bankruptcy on April 2, 2007. In a review of the credit risk of a group of over two hundred REITs and mortgage lenders, we found several firms with high EDF credit measures, which is the one-year probability of default. A look at the EDF credit measure for the group as a whole, however, reveals that credit risk has not changed much for less-risky companies, i.e., those falling within the 75th percentile and below.

Authors: Douglas Dwyer, Ph.D., Sarah WooQu
Date: April 16, 2007

Quantitative rating systems are increasingly being used for the purposes of capital allocation and pricing credits. For these purposes, it is important to validate the accuracy of the probability of default (PD) estimates generated by the rating system and not merely focus on evaluating the discriminatory power of the system. The validation of the accuracy of the PD quantification has been a challenge, fraught with theoretical difficulties (mainly, the impact of correlation) and data issues (eg, the infrequency of default events).

Author: Douglas Dwyer
Date: Spring 2007