Introduction
SAS Risk Engine is a powerful platform in the SAS Viya environment that enables organizations to manage, analyze, and simulate risk across portfolios in a highly scalable, in-memory architecture. It supports advanced capabilities such as scenario analysis, stress testing, and large-scale portfolio computations using modern, cloud-ready infrastructure.
Similarly, the SAS Model Implementation Platform (MIP) provides comparable capabilities within the SAS 9 ecosystem. It allows users to implement, organize, and execute risk models, perform portfolio-level simulations, and manage scenario-based analyses, but within a more traditional SAS 9 architecture.
While the underlying technologies and architectures differ between SAS 9 and Viya, the core functional intent of both platforms remains closely aligned. Both are designed to operationalize risk models and generate analytical outputs for regulatory and business use cases such as IFRS 9, CECL, and stress testing.
An important aspect to note is that many conceptual objects and components in SAS MIP have direct or near-direct counterparts in SAS Risk Engine. For example, model groups, risk data objects, scenarios, and execution flows in MIP often map to pipelines, CAS tables, and execution definitions in Viya.
This conceptual overlap makes prior experience with SAS MIP highly valuable when working with SAS Risk Engine. Understanding these mappings can significantly ease migration efforts, helping teams translate legacy implementations into the Viya framework more efficiently while preserving business logic and analytical intent.
Overview of Object Mapping
The table below provides a high-level mapping of core objects between SAS MIP and SAS Risk Engine. It highlights how key modeling and execution components translate between the two platforms, helping users understand structural equivalence between the platforms. Unsupported objects include risk work groups and user-defined libraries.
Note for Analysis runs:
Direct migration of analysis runs from MIP 3.2 is not supported. You must first publish runs to modeling systems, then import these systems into SAS Risk Engine. Supported runs include:
- Scenario runs – via modeling systems
- Scenario and portfolio cube runs – via modeling systems
- Economic simulation runs – pre-production, via modeling systems
| SAS Model Implementation Platform Object |
SAS Risk Engine Object |
| Analysis run |
Locked risk pipeline |
| Modeling system |
Locked project |
| Method |
Risk method of the same type |
| Model |
Model of the same type (unit tests not supported) |
| Model group |
Model group of the same type (black-box groups not supported) |
| Map |
Risk method map of the same type |
| Pre-execution program |
Program tagged pre-execution |
| Post-execution program |
Program tagged post-execution |
| Time bucket scheme |
Cash flow bucket scheme |
Unsupported runs include portfolio cube runs, model sensitivity analysis, new originations runs, and backtesting runs.
Portfolio Analysis Objects and Pipeline Nodes
This table explains how portfolio-level analytical components in SAS MIP translate into pipeline nodes in SAS Risk Engine. It helps users understand how execution logic is restructured into modular pipeline steps in Viya.
| SAS MIP Portfolio Analysis Object |
SAS Risk Engine Pipeline Node |
| Computed methods |
Query Results |
| Economic simulation data |
Simulate Risk Factors |
| Evaluation model group map |
Evaluate Portfolio |
| Post-execution program |
Execute Custom Code |
| Pre-execution program |
Execute Custom Code |
| Scoring model group map |
Score Counterparties |
SAS MIP Parameters to SAS Risk Engine Pipeline Fields
The following table describes how different parameter types and data inputs in SAS MIP are mapped to corresponding fields within SAS Risk Engine pipelines.
| Parameter Type |
Source Parameter (MIP 3.2) |
Target Pipeline Field (SAS Risk Engine) |
| Input Data |
Counterparty data |
Counterparty data field in Portfolio Data node |
| Economic Simulation Data |
Simulated market states |
Simulated market states data field in Simulate Risk Factors node |
| Mitigation Data |
Mitigation data |
Mitigation data field in Portfolio Data node |
| Portfolio Data |
Simple instrument data |
Simple instrument data field in Portfolio Data node |
| Risk Data Objects |
Risk Data Objects |
Risk Data Object Mappings section in pipeline properties |
| Scenario Economic Data |
SAS libref |
Added to Description field in pipeline properties |
| Advanced Options |
Aggregation levels, cross-classifications |
Saved as _advanced_options program for reference |
| Execution Options: Debugger/Trace |
Debug with trace |
Maps to Trace methods option in Evaluate Portfolio, Score Counterparties, and Query Results nodes |
| Cash Flow Analysis |
Analyze cash flow option, specific cash flow legs |
Cash Flow Bucket Array output variables updated in Evaluate Portfolio node |
| Time Bucket Scheme |
Bucket type |
Mapped to cash flow bucket schemes (All type creates Simple and Cumulative schemes) |
Models and Variable Types
This table compares supported modeling techniques and variable types between the two platforms, highlighting differences in analytical capabilities and limitations.
| Aspect |
SAS MIP 3.2 |
SAS Risk Engine |
| Supported Model Forms |
Cloglog, Code, Constant, Curves, Custom, Fractional Logit, Linear, Logit, Proportional Hazards, Transition Matrix, ASTORE, Enterprise Miner, Python |
Cloglog, Code, Constant, Curves, Custom, Fractional Logit, Linear, Logit, Proportional Hazards, Transition Matrix |
| Unsupported Model Forms |
N/A (all listed forms supported in MIP) |
ASTORE, Enterprise Miner, Python |
| Variable Types |
Bin, Constant, Functional, Piecewise-Linear 1D, Piecewise-Linear 2D, Predefined Root, User-Defined, Variable Expression |
Bin, Constant, Functional, Piecewise-Linear 1D, Predefined Root, User-Defined, Variable Expression |
| Model Unit Tests |
Supported |
Not supported / not migrated |
Model Groups: SAS MIP vs SAS Risk Engine
This table explains how model groups—central to organizing models—are handled differently across the two platforms, particularly in terms of storage, sharing, and limitations.
| Aspect |
SAS MIP 3.2 |
SAS Risk Engine |
| Evaluation Model Groups |
Supported as Evaluation model groups |
Supported as model groups of type Evaluation |
| Scoring Model Groups |
Supported as Scoring model groups |
Supported as model groups of type Scoring |
| Black-box Model Groups |
Supported |
Not supported / not migrated |
| Associated Models |
Models are stored in a global repository and referenced by model groups |
Models are copied into the model group workspace; shared models are part of Shared Models workspace and marked Ready |
| Output Variables |
Defined locally within each model group and method |
Defined globally; conflicts resolved via predefined variables or renaming with method types |
Other Components
This table summarizes additional technical components involved in both the platforms along with possible limitations.
| Component |
SAS MIP (Source) |
SAS Risk Engine (Target) |
Notes / Limitations |
| Transition Matrices |
Defined within model groups |
Defined within model groups |
No major change in structure |
| Risk Data Objects |
Defined with mappings in MIP |
Mappings are preserved via pipeline properties |
Cash Flow types in scoring groups not supported |
| Function Sets |
Stored as SAS datasets |
Stored as CAS tables |
Obfuscated/encrypted code not supported |
| Cash Flow Legs |
Local legs within models |
Converted into global legs |
References updated in output variables |
| User-Defined Logic |
Includes Monte Carlo methods |
Converted to Stochastic methods |
Conceptual mapping, not identical implementation |
| Maps (Evaluation, Mitigation, Scoring) |
Defined with possible black-box references |
Migrated with renaming if duplicates exist |
Black-box references not supported |
| Pre/Post-Execution Programs |
Custom SAS programs |
Program tagged as pre/post execution programs |
May require manual adjustment after migration |
| Time Bucket Schemes |
Defined bucket structures |
Converted to cash flow bucket schemes |
“All” bucket creates both Simple & Cumulative schemes |
Unsupported Features of SAS Model Implementation Platform in SAS Risk Engine
- User-defined libraries
- Model unit tests
- Black-box model groups
A Note on Migration
Migrating from SAS MIP 3.2 to SAS Risk Engine requires careful handling of modeling systems, methods, models, and pipeline configurations. The mip-migration plug-in automates much of the process, but understanding supported objects, mappings, and limitations ensures a smooth migration.
With this guide, risk analysts and modelers can confidently leverage SAS Risk Engine for portfolio analysis, scenario modeling, and risk simulation in the SAS Viya environment.
Conclusion
The transition from SAS Model Implementation Platform to SAS Risk Engine represents more than just a technological upgrade—it reflects a shift toward a more scalable, modular, and cloud-enabled approach to risk analytics. While the core modeling concepts remain consistent, the implementation paradigm evolves significantly, requiring users to adapt to pipeline-based workflows, CAS-driven data handling, and stricter standardization.
Although certain features such as black-box models, user-defined libraries, and unit tests are not carried forward, the enhanced performance, flexibility, and integration capabilities of SAS Risk Engine provide substantial long-term benefits. By understanding object mappings and migration nuances, organizations can effectively preserve their analytical intent while modernizing their risk infrastructure.
Ultimately, familiarity with SAS MIP serves as a strong foundation, enabling a smoother and more intuitive transition to SAS Risk Engine and positioning teams to fully leverage the capabilities of the SAS Viya ecosystem.
Find more articles from SAS Global Enablement and Learning here.
