| Credit Models are intended to aid financial institutions and large corporations in quantifying, aggregating and managing risk across geographical and product lines. |
 |
|
|
| The outputs of these models play increasingly important roles in the risk management process and, for the banks, the setting of their regulatory capital requirements. |
|
| |
|
| Rating entities in the FACT framework |
|
| The rating module in FACT enables you to independently create and manage internal and external credit models for the entire spectrum of your portfolio. |
|
|
|
| You can choose BvDEP databases or alternative sources to drive your models. |
|
|
|
| The module is designed to host a variety of different internal and external models associated to the following elements: |
|
- Companies (standard element in FACT) i.e:
- Banks
- Corporates (private and public)
- Insurance firms
- Retail
- Countries and governmental bodies (standard element in FACT)
- Custom elements
|
|
|
| Our "rating model designer" enables easy implementation of a wide range of models: statistical, score card, structural or hybrid. |
|
|
|
| N.B Integration of third party data sources and methodologies are generally implemented using standard webservice technology. |
|
|
|
| Internal Models |
|
| FACT's “model designer" is a unique component that enables you to create, modify and enhance your internal rating models, with or without Bureau van Dijk's assistance. |
|
|
|
| The module has been built to allow users with no specific programming skills to implement a full rating model independently. Using contextual menus, users can choose between different types of item evaluation (quantitative or qualitative) to migrate models from, for example, an Excel Spreadsheet into FACT. |
|
|
|
| If certain methodologies require more complex computations such as statistics or probabilities, a Visual Basic / C# editor is also available. This enables the user to structure the model in the same way as a typical Excel macro. |
|
|
|
| As illustrated in the figure below, the implementation of a model begins with a detailed analysis of the methodology. |
|
|
|
| The model architect implements the model using FACT’s wizard driven ‘model designer’. Following testing, the model is then released into the live FACT environment for use by the analysts. |
|
|
|
| Model Designer |
|
|
|
| The model designer empowers specific users to create and edit an unlimited number of different models, each with their own administrative profile. |
|
|
|
ENLARGE IMAGE
|
|
|
|
| Functionalities available |
|
- Ability to include qualitative as well as quantitative factors within the same rate card. Examples of qualitative factors that can be included: industry codes, event risk, financial and performance, management, public, private and country rating benchmarks
- Models can be created, copied, modified, validated and deleted in a parallel test environment.
- All internal models are stored and locked in a central data repository.
- The results of a model can vary from a rating to a score, PD or LGD.
- Ability to create multiple display layouts: e.g. year formats, currencies and exchange rates.
- Easily configured to produce multiple borrower risk rating statuses (preliminary, indicative, proposed, approved etc).
- Flexibility to change financial drivers to produce alternate ratings ['what if analysis' etc].
- Ability to attach documents to the obligor or archived rating / default probability.
- Ability to assign specific models to obligors based on business segment or standard industry code.
|
|
|
|
|
Algo-CRS Models |
|
| FACT allows you to automatically apply the rating models from Algorithmics to entities included in BvDEP databases or alternative data sources.(The entities can be public or private companies or financial institutions). |
|
|
|
| The models can be applied to both historic and financial statement data and current market information. They will generate statistically robust, 'agency-like' ratings for organizations where such evaluations are simply not available. The models produce through-the-cycle ratings that are within two notches of major agency ratings approximately 86% of the time for listed companies, and 81% of the time for private companies |
|
|
|
| The following models are available: |
|
|
|
| Corporate and Utilities rating models |
|
| Corporate and Utilities rating models are built and tested using ratings for major North American, Western European, Japanese and Australian corporations with assets in excess of 50 million USD. |
|
| |
|
| Non Investment Grade Models |
|
| Non Investment Grade Models utilize the same methodology as the general corporate rating models, calibrated and optimized for organizations known to be non-investment grade, such as small to medium enterprises. |
|
| |
|
| Bank Rating Models |
|
| Seven different bank models are available, each adjusted for a particular country or region. They are designed for regulated commercial banks whose primary business is savings and loans or commercial lending. For additional integrity they are benchmarked against a combination of Fitch Individual Ratings and the equivalent, publicly available ratings from other major agencies. |
|
| |
|
|
|
| Why use the CRS rating models |
|
- CRS models can serve as the basis of any credit approval process of a bank or corporate that incurs credit risk.
- The CRS rating model is a fast and reliable way for senior management or audit teams to benchmark and validate the performance and consistency of internally developed models or ratings.
- The models will generate agency like ratings when they are not available either for public of private companies or perform “what if" scenarios for publicly rated firms.
- No judgemental input is applied; therefore a portfolio of names can be run through a consistent and standard model, obtaining an independent and unbiased rating.
- Non financially trained employees can use the model to rate a counterparty and rapidly assess its creditworthiness.
- Vendor financing or commercial terms can be based upon model results.
|
|
|
| Corporate Suite |
|
| A suite of rating models that produces agency like long term ratings of the standalone financial strength of private and public corporates. Model prediction performance is measured by assessing how well the model predicts agency rating. |
|
|
|
| These quantitative models generate an agency like issuer credit rating for an industrial company or utility. (Each model is specific for its sector). The model assesses the standalone creditworthiness of the company. They do not take account of any governmental or parental support when generating a credit assessment. |
|
|
|
| The models rating is purely derived from the quantitative non-subjective financial statement and equity price derived inputs. The models generate issuer like ratings linked to a company’s likelihood of default. |
|
|
|
| N.B The ratings do not include any assessment of structures or likely recovery in the event of default. |
|
|
|
| The model ratings and scores are driven by the company’s most recent financial statements and, if available, equity price and volatility. Therefore, in the case of the Corporate Suite (General Corporates, Non-Investment Grade and Utilities) there are two models: public and private. |
|
|
|
| For General Corporates, there is a model for each of the three regions North America, Western Europe and the Rest of the World. |
|
|
|
| For Non-Investment Grade, the model is applied globally and in the case of Utilities the model is applied to Developed Markets only. In each case, the public model can be used to assess companies which have publicly traded equity. |
|
|
|
| The private model does not require equity information and is suitable for assessing any company, publicly or privately owned, using the latest financial statements only. |
|
|
|
| The models are built by capturing the statistical relationships between a company’s financial fundamentals and rating using at least seven years of data for large samples of rated companies. The models in effect perform a massive benchmarking exercise enabling an analyst to instantly assess where any company should sit on the credit scale of AAA to C (BBB- to C for NIG) based on the financial fundamentals. |
|
|
|
| Methodology - Rating |
|
| The rating used to calibrate the models is a composite of the published Fitch, Moody’s and/or Standard & Poor’s ratings. |
|
|
|
| Methodology - Equity Data for public companies, equity market data is also used to produce a simple distance to default credit measure. Equity data provides a more forward looking perception of the performance of a company and incorporates qualitative factors. Therefore, it complements information provided by financial statements and provides a better fit for the model. |
|
|
|
| Model Build Method |
|
| The CRS models are statistically based quantitative models based on Ordered Logistic Regression. |
|
|
|
| CRS corporate rating models use both historic financial statement data and, where applicable, current equity market information to generate statistically robust, “agency-likeö ratings for both public and private firms. |
|
|
|
ENLARGE IMAGE
|
|
|
|
| Bank Suite |
|
| A suite of rating models that produce agency like long term ratings of the standalone financial strength of a bank, as well as the log term ratings which incorporate government support. |
|
|
|
| The Algo CRS bank suite contains seven regional bank rating replication models, each designed to rate banks in a particular geographic region. (Simply using one bank model for all regions is not appropriate and does not effectively replicate agency ratings). |
|
|
|
| Bank models vary by region for two main reasons: |
|
|
|
| (1) Variations in the regulatory and economic environment mean the criteria used by agencies to rate banks vary. |
|
|
|
| (2) Differences in accounting standards mean there are different levels of financial detail available. For example, in the United States Algorithmics have access to greater financial detail about the composition of assets and liabilities than in Eastern Europe. |
|
|
|
| Algorithmics build models that are benchmarked using Fitch Individual Ratings. These ratings are based on the bank’s financial fundamentals and do not take into account the probability that the bank will receive external governmental or parental support, nor do they address risks arising from sovereign actions that may interfere with a bank’s ability to honor its domestic or foreign currency obligations. |
|
|
|
| The Fitch Individual Rating (FIR) scale is: |
|
|
|
|
|
|
|
| Note that the model rating is derived purely from the quantitative non-subjective financial statements. It does not incorporate any subjective analyst input, other than for the estimate of support. |
|
|
|
| Methodology - Rating |
|
| There are two types of ratings generated by the models, the Individual Rating which is based solely on the individual financial strength of the bank and the Long Term Ratings which takes into account external support. |
|
| |
|
| The individual ratings used to calibrate the models are the published Fitch Individual Ratings three months after the financial statement date. The reason for the time lag is that financials are published subsequent to the period end. Taking the rating at the exact statement date simply wouldn’t correspond to the new data and hence the relationship would look weaker. For the Long Term Ratings, Algorithmics use the LTR and government support ratings published by Fitch. |
|
|
|
| Methodology - Data |
|
| Financial data comes from a global database, namely “Bankscope", published and distributed by BvDEP. The data consists of consolidated/unconsolidated balance sheet and income statement items totaling up to 200 different data items. N.B. Equity data is not used for bank models. |
|
|
|
| Model Build Method |
|
| The models are statistically based quantitative models based on Ordered Logistic Regression. |
|
|
|
| Algorithmics start with a pool of 74 financial items and ratios, grouped into seven clusters according to nature of the quantity they measure. |
|
|
|
| __________________________________ |
|
|
|
| EIU Country Rating System |
|
| A fully integrated EIU country risk model enables users analyze key emerging markets. |
|
| The model can be used as your institution’s primary tool for analyzing country credit risk, as input into your in-house risk assessment process or as a check on your institution’s own country assessments. The web-based risk model assesses the country credit risk of 100 emerging markets and highly indebted countries and provides an early warning system of financial crises in developing countries. Ratings and scores for six regional aggregates determine the riskiness of a region and serve as a benchmark for comparison. |
|
| What does the Risk Model provide you with? |
|
| A comprehensive service covering 100 countries and six regions based on 77 indicators, both quantitative and qualitative, ranging across 13 different risk categories. |
|
| Timely and accurate monthly risk assessments by our team of over 80 country risk analysts |
|
| Access to a database unequalled by any other service available in the marketplace, with scores and ratings going back to 1997. |
|
|
|
| What functionality does the Risk Model provide? |
|
- Adjust the weightings and scores of any of the individual 77 indicators or choose a combination of our indicators and your own to derive a customised set of risk scores.
- Compare the indicator scores across countries, as well as view the historical scores for a single country.
- Access the data with simple to use navigation.
- Export and chart data by date, by country or ratings category.
|
|
|
| What risk categories are provided by the Risk Model? |
|
- Current account
- Debt structure
- Exchange-rate policy
- Liquidity
- Political efficacy
- Global climate
- Growth / savings
- Monetary policy
- Fiscal policy
- Political stability
- Regulatory policy
- Trade policy
|
|
|
| How the EIU compare with ratings agencies? |
|
| Unlike the ratings agencies, EIU give risk assessments via the model across 100 countries every month. The result is they catch deteriorating or improving trends often before the ratings agencies issue formal warnings. |
|
| As part of the well-respected Economist Group, the EIU provide ratings that are totally independent. |
