SAS Documentation
SAS® Solution for Stress Testing
Reference manual - version 08.2021
Loading...
Searching...
No Matches
Introduction

Introduction

SAS® Solution for Stress Testing addresses the needs of financial institutions to conduct rigorous stress testing activities to meet regulatory expectations. There are two primary types of stress tests that must be performed:

  • regulatory stress tests: These are required external stress tests that are imposed by third-party regulatory agencies.
  • "business as usual" stress tests: These are self-imposed internal stress tests that can be used to plan for and quantify risks resulting from strategic changes to an organization's portfolio over time.

Both types of stress tests require that financial institutions calculate potential credit losses and create loss reserves in response. The solution enables users to perform both types of stress testing by providing the following:

  • robust data quality validation
  • strong analytical model development capabilities
  • a solid control and governance environment
  • a seamless platform for risk and finance collaboration
  • a process workflow to orchestrate stress testing activities between relevant parties

The result is a holistic view of an organization's balance sheet, income statement, liquidity positions and capital requirements.

The stress testing cycle takes initial inputs (financial statements, portfolios, growth assumptions, scenarios, and so on) through a process of calculation and data aggregation resulting in projected financial statement reports, as shown in detail in the following diagram.


SAS Solution for Stress Testing supports the following basic process (that is, the stress testing cycle):

  1. Data at time t0 is placed into a landing area from existing financial record keeping systems and initially populated.
  2. The data is prepared, processed, and reviewed to ensure it meets data quality standards for stress testing.
  3. Master risk scenarios and business evolution plans are configured in parallel to account for risk and growth assumptions.
  4. Stress testing and economic capital models are then run in parallel and results are reported, reviewed, and adjusted as needed.
  5. For stress testing models, projected financial statements (balance sheet projections) are generated for future periods (t1 to tn).
  6. Projected financial statements are reviewed and approved, at which point management actions can be applied to the financial statement data.
  7. After the stress test and economic capital results are reviewed and deemed adequate, sign-off occurs.


SAS Solution for Stress Testing Framework

The following figure provides a high level overview of the architectural framework of this solution.

Data Management, Data Quality, and Overrides

SAS Solution for Stress Testing provides the following features as it relates to the management of data:

  • pre-built input and output data model blueprints for ICAAP and EBA exercises, with easy to configure data definition schemas
  • easily configured pre-defined rules and related actions that are mapped to BCBS239 principles (accuracy, integrity, completeness, and so on) and can operate on individual or aggregated data to perform quality checks
  • network diagrams that enable you to track usage, determine data lineage, and understand data associations across stress testing activities
  • an analysis data inventory that aggregates all analysis data types (portfolios, data quality results, financial information, and so on), enables analysis of individual records, and enables you to apply business and allocation rules to modify and adjust (override) the data

Growth Assumptions and Scenario Design

SAS Solution for Stress Testing has the following features as they relate to scenario design:

  • segmentation schemes for defining business segments for relevant portfolios, according to the requirements of a given stress testing exercise
  • allocation schemes that enable you to project your segment-by-segment portfolio mix
  • business evolution plans that utilize segmentation and/or allocation schemes to project the front book at the required level of granularity
  • a centralized scenario repository manager that enables you to define scenarios, load existing regulatory scenarios, change or enrich scenarios with relevant risk factor data, apply scenarios to selected portfolios, and explore scenario results to understand the potential impacts
  • an inventory of master risk scenarios for exploring and comparing risks under different levels of severity

Execution Engine and Simulation Environment

SAS Solution for Stress Testing contains the following features pertaining to model execution and simulation:

  • a front-book generation engine that projects the portfolio based on business outlook assumptions (e.g. volume growth, or new production growth per portfolio segment), thereby enabling an organization to use static, constant or dynamic balance sheet assumptions
  • a centralized model inventory that keeps track of all the models, scripts and versions. Models can be stored from the institution’s own model library, thus providing an audit trail for all models used within the stress test process
  • support for the configuration of parameters that can be created, edited, categorized, and called by any model run to support the quantification of impacts of stress scenarios
  • the ability to run models on a stand-alone basis or as part of a model group that incorporates dependencies and aggregation logic through the configuration of user-defined logic (UDL)
  • an efficient execution process that can make calls to SAS Model Implementation Platform for the high-performance execution of SAS scripts or other model types
  • a robust calculation infrastructure that promotes an efficient simulation environment and supports multiple executions of different assumptions and inputs
  • the ability to perform what-if analyses and define relative or absolute variations on risk factors and portfolio attributes
  • pre-built methodology templates for expected credit loss and risk-weighted asset (RWA) forecasting
  • an extensible environment for creating and dropping in other custom or pre-built templates (including NII, PPNR, Market Risk, Operational Risk, Conduct Risk, and other risk forecasting templates that are in plan for future releases)

Orchestration and Risk Consolidation

SAS Solution for Stress Testing provides the following orchestration and risk consolidation capabilities:

  • a configurable analysis run environment that orchestrates the timely execution of a combination of models, master risk scenarios, and business projections
  • a sample workflow that maps all relevant activities and assigns individual tasks to respective organizational units
  • task-level status tracking capabilities for monitoring the project and identifying performance issues

Dynamic Reporting

SAS Solution for Stress Testing has the following dashboarding and reporting features:

  • pre-built SAS Visual Analytics reports to display overall data quality metrics based on the individual results obtained for each data quality rule and the weight assigned to each rule
  • drill-down capabilities to perform root-cause analysis of data quality issues, track data quality issues to individual records, and remediate and correct data quality issues



Products Involved and Their Various Roles

SAS Solution for Stress Testing combines a powerful and integrated set of SAS products and solutions to help financial institutions successfully conduct their stress testing activities.

SAS Risk Governance Framework (RGF)

SAS Solution for Stress Testing leverages SAS Risk Governance Framework to:

  • provide a common interface to perform end-to-end stress testing activities
  • control and govern centralized workflows that orchestrate the stress testing process and provide complete auditability and traceability
  • provide a transparent platform for processing large volumes of data, performing transformations on data, initiating analysis runs, and recording data lineage using relationships
  • provide a robust documentation management platform with version controls and user entitlements

SAS Model Implementation Platform (MIP)

SAS Solution for Stress Testing leverages SAS Model Implementation Platform to:

  • provide a controlled environment where complex systems of loan-level risk models can be implemented quickly and transparently
  • offer full integration with both SAS Risk Governance Framework and SAS Infrastructure for Risk Management for high-performance analytics

SAS Infrastructure for Risk Management (IRM)

SAS Solution for Stress Testing leverages SAS Infrastructure for Risk Management to:

  • Provide high-performance analytics and data manipulations
  • Leverage in-memory analytical capabilities

Other Platform Components

In addition to the three platform components described above, SAS Solution for Stress Testing includes other products from SAS risk portfolio including the following:

  • SAS Risk Scenario Manager (RSM)
  • SAS Visual Analytics (VA)

Here is a visualization of the various components showing how the products interoperate within the solution.


Solution Architecture

The solution architecture is designed to cover an end-to-end stress testing cycle that consists of the following tasks:

  1. Load the necessary data into the system.
  2. Process the data through an orchestrated and governed process that includes analytical transformations.
  3. Initiate model runs and automatically create records of all input and output objects that were used in the analysis.
  4. Create reports and publish results.

Landing Data

The landing area (LA) is the input data mart of the SAS Solution for Stress Testing. The landing area can be any database or data warehouse. The default solution landing area is a folder on the server that contains data in a SAS data set format. It contains subsets of the input data (depending on which analysis periods were chosen). The tables are versioned by date (MMDDYYYY).

SAS Solution for Stress Testing contains a number of default tables that are necessary to run the credit risk stress testing analytical model that ships with the solution and to perform balance sheet projection. The table below lists the input data in the landing area. For every analysis cycle, these tables must be loaded to the landing area prior to registering the tables in the solution.

Table Description
CASHFLOW Contains cashflow data for instruments in the credit portfolio table.
COLLATERAL Collateral data used for calculation of economic capital
COUNTERPARTY Contains counterparty data.
CREDIT_PORTFOLIO Contains detailed information about credit instruments.
CREDIT_SYNTH_PORTFOLIO Contains sample data for use in generation of synthetic instruments.
ECONOMIC_SIM Economic data for simulation based EC calculation
ENTITY This table is used to provide details about business entities.
GL_ACCOUNT Contains general ledger account data.
INDIVIDUAL_ADJUSTMENT Instrument level movement data for adjustments
MARKET_COUNTERPARTY Contains market counterpaty data.
MARKET_ECONOMIC_SIM Market economic data for simulation based EC calculation
MARKET_PORTFOLIO Contains detailed information about market instruments.
MITIGATION Mitigants data for use in economic capital calculation
OWN_FUNDS_ACCOUNT Contains own funds account data.
PROFIT_LOSS_ACCOUNT Contains profit and loss account data.
RWA_WEIGHTS Contains RWA weights data for balance sheet assets used in financial statement projection
SYNTHETIC_CASHFLOW Cashflow data for instruments in the credit portfolio table

Analysis Data
Data in the landing area is registered in the solution during the data preparation run in the cycle or by executing an Analysis Run of the default ST- Data Preparation model. In both cases the analysis run creates a data definition that captures the structure of the data in the Data Definition object and registers the data in an Analysis Data object. Users can view the detail data in the solution through this Analysis Data object. The analysis data can be used directly in various analysis runs and model runs.

Output Data
There are two areas where output data is stored:

  • In-memory report tables: These tables that contain a limited volume of output data and exist only to service the SAS Visual Analytics reports.
  • EWST data repository: These tables contain input data from landing area and any output data created from analysis, and serve as the data model for the solution. Further details on the EWST data repository are provided in FAQ: What is the Enterprise Wide Stress Testing data repository?

Additional Documentation
For additional information around these products and their respective capabilities, please refer to their respective user and administration manuals on the SAS support website.

What's New in Stress Testing 08.2021


The 08.2021 release of Stress Testing creates a new perspective: Comprehensive Capital Analysis and Review (CCAR).

  • New Perspective for the Comprehensive Capital Analysis and Review (CCAR)
    • New workflow template for expanding the portfolio and calculating Credit Risk Measures
    • Data preparation, Segment-Level model, Portfolio Expansion, Credit Risk Measures model, review of results, and signoff, with sample cycles for multiple entities
    • Signoff model that allows results to be tagged for use in production runs in the aggregation process (described below)
    • CCAR Report Templates
      • Sumary Totals
      • Expected Loss Amount by Rating
      • NCO Summary
      • Expected Loss by Exposure at Default
      • Risk Rating Loss Rate by Starting Risk Rating
      • EAD Migration Summary
      • Incremental EL by Product
  • ICAAP Market Risk
    • New Market Risk Short & Long Term analysis based on simulations for Economic Capital and Value-at-Risk models
    • New Market Risk Short Term model
    • New ICAAP Market Risk sample cycles
    • New Market Risk Short and Long Term reports
  • Risk Stratum 03.2021 Enabled Enhancements
    • Control Framework
    • Sample Python models
    • Process Dashboard
  • New ICAAP report templates

    • Capital Required
    • Capital Adequacy

What's New in Stress Testing 06.2020


The 06.2020 release of Stress Testing separates the solution into two perspectives: Enterprise Wide Stress Testing (EWST) and Internal Capital Adequacy Assessment Process (ICAAP). The STM perspective is no longer used. Functionality that was in the STM perspective has been divided amongst these two, with additional functionality added as well.

  • New Perspective for the Internal Capital Adequacy Assessment Process (ICAAP)
    • New workflow template for calculating Economic Capital
      • Data preparation, Economic Capital models, review of results, and signoff, with sample cycles for multiple entities with Credit, Market, and Operational risk
      • Signoff model that allows results to be tagged for use in production runs in the aggregation process (described below)
      • Ability to import externally calculated Economic Capital and VaR results
    • Risk Aggregation
      • New workflow template for risk aggregation
      • Aggregation model that takes Economic Capital for different risk types across entities and calculates an aggregate VaR and Economic Capital
        • Linear aggregation model
        • Hierarchical COPULA aggregation model that supports several COPULA models
    • ICAAP reporting templates
      • Summary
      • Scenario Details
      • Adjustments
      • Concentration Risks
      • Balance Sheet and Income Statement
      • Management Actions Adjustments
      • Capital Ratios
  • New Perspective for Enterprise Wide Stress Testing (EWST)
    • Simplified workflow with Economic Capital removed (moved to the new ICAAP perspective)
    • Integration with ICAAP perspective so that Portfolio data, Scenarios, Business Evolution Plans, and other objects can be shared easily
    • Integration of Stress Credit Risk and Balance Sheet Projection results with ICAAP reports

What's New in Stress Testing 03.2020


The 03.2020 release of Stress Testing moved the solution to the new SAS Risk Stratum framework, an integrated platform that allows for coexistence of solutions and the sharing of objects between solutions. This release does not include new functionality specific to Stress Testing, but does include the improvements provided by the new platform.

For more information about supported products for this release, see Compatibility.

What's New in Stress Testing 09.2019


  • New Model Template for Stress ECL and RWA (EBA 2020)
    • This methodology is similar to the methodology used for the EBA 2018 guidelines. However, for EBA 2020 guidelines, the mean reversion of the ECL is obtained by using a dynamic average of the the ECL calculated by the base case and severe scenarios.
    • This dynamic average progressively gives more weight to the base case scenario. Both scenarios, base case and severe, do not revert to the mean after the forecast time. Instead, they remain in "stressed" condition.
  • New Economic Capital Model Templates – Credit Risk+ and Simulation Based
    • Two model templates have been added to calculate economic capital (EC) and credit risk Value-at-Risk (VaR) using two different methodologies:
      • Credit Risk+ Method: The evaluation of the EC and VaR mainly consists of approximating the compound loss distribution using the "Convolution with Fast Fourier Transform" technique.
      • Monte Carlo Simulation-Based Method: This template uses simulation to address the strategic risk that financial institutions face on a daily basis. This model template is used to evaluate EC and VaR.
    • Note: These two methodologies are an adaptation of EC and VaR calculation methods from the SAS Risk Management for Banking credit risk product, which is no longer available for sale.
  • Data Aggregation and Balance Sheet Population Rules Template
    • Aggregate credit risk analysis is performed at the instrument level and applied to line items on the balance sheet by defining mapping rules.
    • The template enables analysts to define volume, yield and value growth assumptions for balance sheet and P&L line items.
    • These assumptions can then be used to project balance sheet, P&L, and own funds line items into the future at both the portfolio and aggregate level.
  • Management Actions Rule Sets
    • Users can now adjust financial statement projections by defining management actions that make adjustments to different line items for a given management action.
  • Process Narratives and CAP Assessments
    • Users can set up regulatory expectations and conduct self-assessments of expectations for the effective capital adequacy planning (CAP) process.
    • In addition, users can create and manage narratives for each stress testing activity to support regulatory expectations of the internal capital planning process.
  • New Management Reports
    • New management reports have been added to the reporting dashboard in support of financial statement projection
    • New management reports have been added to the reporting dashboard in support of the Credit Risk VaR and Economic Capital calculations.
  • Updated Cycle Workflow to Support Balance Sheet Projection, Economic Capital Analysis, and Management Actions
    • The Cycle object workflow now displays new tasks for configuring balance sheet projection inputs, performing balance sheet projection, calculating economic capital, and applying management actions.
    • The Cycle workflow now supports the creation of new iterations of the cycle from any task in the cycle.
  • Batch Execution of Cycle Activities
    • You can now automate the execution of all tasks in the cycle from end-to-end in batch mode without having to click through the UI. The Batch Execution Service runs through the entire cycle (or part of the cycle depending on user preference).

For more information about supported products for this release, see Compatibility.

Disclaimer

SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration. Other brand and product names are registered trademarks or trademarks of their respective companies.