NUS-RMI Credit Research Initiative Technical Report
Version: 2013 Update 2b
RMI staff article
For any questions or comments on this article, please contact Oliver Chen at
[email protected] Keywords: Non-profit credit research initiative, credit risk, probability of default, forward intensity.
This document describes the implementation of the system which the Credit Research Initiative (CRI) at the Risk Management Institute (RMI) of the National University of Singapore (NUS) uses to produce probabilities of default (PDs). As of this version of the Technical Report, RMI covers around 60,400 listed firms (including delisted ones) in 106 economies around the world (see Table A.1). Of the over 39,000 active firms under the CRI coverage, around 34,000 firms have sufficient data to release daily updated PDs. The PD for all firms is freely available to users who can provide evidence of their professional qualifications to ensure that they will not misuse the data. General users who do not request global access are restricted to a list of 3,000 firms. The individual company PD data, along with aggregate PDs at the economy and sector level, can be accessed at http://rmicri.org.
The primary goal of this initiative is to drive research and development in the critical area of credit rating systems. As such, a transparent methodology is essential to this initiative. Having the details of the methodology available to everybody means that there is a base from which suggestions and improvements can be made. The objective of this Technical Report is to provide a full exposition of the CRI system. Readers of this document who have access to the necessary data and who have a sufficient level of technical expertise will be able to implement a similar system on their own. For a full exposition of the conceptual framework of the CRI, see Duan and Van Laere (2012).
The system used by the CRI will evolve as new innovations and enhancements are applied. The changes to the 2012 technical report and operational implementation of our model are: (1) RMI’s global coverage; (2) extension of the forecast horizon to five years by applying a Nelson-Siegel type parameterization and using a Sequential Monte Carlo (SMC) method; (3) changes to treatment of financial statements in monthly calibration; (4) exclusion rule for merger and acquisition (M&A) events; (5) changes to treatment of companies after a default event; (6) change from monthly sigma to daily sigma; (7) changes in the treatment of missing values; and (8) changes in the level and trend calculations and (9) changes in the Distance-to-Default computation. This version of the technical report provides an update on the operational implementation of the CRI and includes all changes to the system that had been implemented by September 2013. More specifically, in addition to Version: 2013 update 1, the current version of the technical report specifies some revisions to the monthly parameter updates that went into effect as of the August 2013 calibration. The latest version of the Technical Report and addenda to the latest version are available via the web portal and will include any changes to the system that have been implemented since the publication of this version.
The remainder of this Technical Report is organized as follows. The next section describes the quantitative model that is currently used to compute PDs from the CRI. The model was first described in Duan et al. (2012). The description includes calibration procedures, which are performed on a monthly basis, and individual firm PD computations, which are performed on a daily basis.
Section 2 describes the input variables of the model as well as the data used to produce the variables for input into the model. This model uses both input variables that are common to all firms in an economy and input variables that are firm-specific. Another critical component when calibrating a probability of default estimation system is the default data, and this is also described in this section.
While Section 1 provides a broader description of the model, Section 3 describes the implementation details that are necessary for application, given real world issues of, for example, bad or missing data. The specific technical details needed to develop an operational system are also given, including details on the monthly calibration, daily computation of individual firm PDs and aggregation of the individual firm PDs. Distance-to default (DTD) in a Merton-type model is one of the firm-specific variables. The calculation for DTD is not the standard one, and has been modified to allow a meaningful computation of the DTD for financial firms. While most academic studies on default prediction exclude financial firms from consideration, it is important to include them given that the financial sector is a critical component in every economy. The calculation for DTD is detailed in this section.
Section 4 shows an empirical analysis for those economies that are currently covered. While the analysis shows excellent results in several economies, there is room for improvement in a few others. This is because, at the CRI’s current stage of development, the economies all use the variables used in the academic study of US firms in Duan et al. (2012). Future development within the CRI will deal with variable selection specific to different economies, and the performance is then expected to improve. Other planned developments are discussed in Section 5.
I. MODEL DESCRIPTION
The quantitative model that is currently being used by the CRI is a forward intensity model that was introduced in Duan et al. (2012). Certain aspects of the model are taken from Duan and Fulop (2013). This model allows PD forecasts to be made at a range of horizons. In the current CRI implementation of this model, PDs are forecasted from a horizon of one month up to a horizon of five years. At the RMI CRI website, for every firm, the probabilities of that firm defaulting within one month, three months, six months, one year, two years, three years and five years are given. The ability to assess credit quality for different horizons is a useful tool for risk management, credit portfolio management, policy setting and regulatory purposes, since short- and long-term credit risk profiles can differ greatly depending on a firm’s liquidity, debt structures and other factors.
The forward intensity model is a reduced form model in which the PD is computed as a function of different input variables. These can be firm-specific or common to all firms within an economy. The other category of the default prediction model is the structural model, whereby the corporate structure of a firm is modeled in order to assess the firm’s PD.
A similar reduced form model by Duffie et al. (2007) relies on modeling the time series ...