In this blog we will be discussing how financial attributes impact your expected credit loss. When it comes to calculating expected credit loss (ECL), an attribution template plays a crucial role in ensuring accuracy and efficiency. An attribution template defines the objects used in an attribution analysis, including input data definitions, output variables, and attribution factors. This structured approach allows for a comprehensive analysis of the factors influencing ECL. Attribution analysis enables you to gain an understanding of what factors contributed to your period-over-period changes in expected credit losses. For example, if you estimated the expected credit loss last quarter and this quarter, you can use attribution analysis to determine how changes to the portfolio, models, and scenarios have individually contributed to the overall change in the expected credit loss. You could then answer the following: How much has the expected credit loss changed due to newly acquired positions? What was the impact of the changes to our projected GDP growth on ECL? How have the  new credit models impacted our ECL estimates? The attribution analysis makes incremental changes to the analysis performed in a previous period and records the values of user-specified output variables with each change. Specifically, the previous period's analysis is altered by replacing components with components from the analysis performed in the current period. Attribution analysis enables customers to: create waterfall reports detailing how changes in an analysis run's inputs have impacted the outputs. create multiple Attribution Templates to streamline the creation of these reports. quickly and easily run an attribution analysis using an Attribution Template, as part of a Cycle or as a separate Analysis Run. compare any two runs of the same analysis model, whether the underlying models are MIP models, Python models, R models, or SAS scripts. Understanding Attribution Templates An attribution template is essentially a blueprint that outlines the components required for an attribution analysis. It includes: Input Data Definition: This specifies the data sets and variables that will be used in the analysis. It ensures that all relevant data is included and properly formatted for the analysis. Output Variables: These are the variables that will be tracked and analyzed during the attribution process. They represent the key metrics that will be used to assess changes in ECL. Attribution Factors: These are the specific factors that will be analyzed to determine their impact on ECL. The attribution template typically includes multiple factors, such as expired positions, portfolio aging, new originations, and reclassifications.   The Role of Attribution Factors Attribution factors are the core elements of the attribution analysis. They represent the various components that can influence changes in ECL. For example, the template might include factors such as: Expired Positions: Analyzing the impact of expired positions from one period to the next. Portfolio Aging: Assessing how the aging of the portfolio affects ECL. New Originations: Evaluating the impact of new loan originations on ECL. Reclassifications: Understanding how reclassifications within the portfolio influence ECL. Conducting an Attribution Analysis The attribution analysis is conducted by running the analysis repeatedly, with each run focusing on a different attribution factor. The order in which the factors are analyzed can affect the results, as each factor's impact is assessed in the context of the others. For example, the analysis might first evaluate the impact of expired positions, then move on to portfolio aging, and so on.   Creating and Modifying an Attribution Template Creating an attribution template involves several steps: Define Input Data: Specify the data sets and variables that will be used in the analysis. Select Output Variables: Identify the key metrics that will be tracked and analyzed. Choose Attribution Factors: Determine the specific factors that will be included in the analysis. Modifying an attribution template is an ongoing process. As new data becomes available or as the business environment changes, the template can be updated to reflect these changes. This ensures that the ECL calculations remain accurate and relevant. Conclusion An attribution template is a powerful tool for supporting the calculation of expected credit loss. By providing a structured framework for analyzing the factors that influence ECL, it helps organizations make more informed decisions and better anticipate potential losses. Incorporating both quantitative and qualitative factors into the attribution analysis ensures a comprehensive and realistic assessment of credit risk.   ... View more

The purpose of this blog is to discuss the importance of using qualitative factors if adjustments to the calculated expected  loss is necessary. When calculating expected credit loss (ECL), it's crucial to incorporate both quantitative and qualitative factors to ensure a comprehensive and accurate assessment. While quantitative factors rely on historical data and statistical models, qualitative factors provide additional context that can significantly influence the final ECL estimate.   Understanding Qualitative Factors Qualitative factors, often referred to as Q-Factors, are adjustments made to the quantitative ECL model to reflect current and future conditions that may not be captured by historical data alone. These adjustments allow management to incorporate economic, business, and other relevant factors into their credit loss calculations.   Key Qualitative Factors Economic and Business Conditions: Changes in the economic environment, such as recessions or booms, can impact borrowers' ability to repay loans. Adjusting for these conditions helps in aligning the ECL with the current economic reality. Management's Experience and Judgment: The experience and judgment of lending management play a crucial role in assessing credit risk. Factors such as the depth of management's experience and their ability to navigate challenging conditions are considered. Portfolio Characteristics: The nature and volume of the loan portfolio, including the terms of loans and the types of borrowers, can influence credit risk. For instance, a portfolio heavily concentrated in a particular industry may require adjustments based on industry-specific risks. Collateral Value: The value of collateral securing loans can fluctuate, affecting the potential recovery in case of default. Adjustments for changes in collateral value ensure that the ECL reflects the true risk of loss. Credit Concentrations: The existence and effect of credit concentrations, such as loans to a single borrower or industry, are important qualitative factors. High concentrations can increase risk and necessitate adjustments. Applying Qualitative Adjustments To apply qualitative adjustments, analysts typically create a Q-Factor rule set that defines how each qualitative factor will impact the ECL. This rule set is then applied to the quantitative model to adjust the ECL estimates. For example, adjustments might be made based on geographic location, instrument type, or other attribute-based filters. The SAS Allowance for Credit Loss solution provides a sample Q-Factor adjustments rule set. Each factor can be assigned a specific sentiment by the quantitative analyst. There are five customizable options. Positive and limited positive impact, which would result in a decrease of ECL. No Impact means no change in ECL. And a Negative and limited negative impact would increase the ECL. An analyst could review a rule factor and determine that the economic and business conditions that affect the collectability of the portfolio could have a limited negative impact on the expected credit loss.   Modifying a Qualitative Factor Adjustment Rule Set The process of modifying a qualitative factor adjustment rule set involves several steps: Creating a Q-Factor Rule Set: Analysts start by creating a Q-Factor rule set that includes various qualitative factors and their corresponding adjustments. This rule set can be applied to specific groups of instruments using attribute-based filters such as geography or instrument type. Applying Weighted Adjustments: Each instrument within the defined filter is assigned a weighted adjustment based on the Q-Factor rule set. This ensures that the ECL reflects both current and future conditions. Updating the Rule Set: The rule set can be updated as needed to reflect changes in economic conditions, portfolio characteristics, or other relevant factors. This allows for continuous improvement and accuracy in the ECL calculation. Conclusion Incorporating qualitative factors into the calculation of expected credit loss is essential for a realistic and comprehensive assessment of credit risk. By considering both quantitative data and qualitative insights, organizations can better anticipate potential losses and make more informed lending decisions.   ... View more

  Allocation rules are used during the adjustment task in the workflow within SAS Allowance for Credit Loss. After reviewing the output of the Credit Risk Analysis step, the quantitative analyst might want to apply an adjustment and review the impact on the expected credit loss. you can leverage rule sets to apply adjustments.   From the Rule Set workspace, the quantitative analyst can create an allocation rule set and then apply it from the workflow. ​ ​This approach enables the analyst to define​:   the adjustment value​ the adjustment type​ how it is allocated​ and how it is aggregated.   When defining an allocation rule set, you will need to determine the adjustment rule method.   The Adjustment Value Rule Method, is the method in which the adjustment value will be applied. You can use the Value method where the value of the adjustment is considered as a constant number that is used by the rule. Or you can use the Formula method where the adjustment is considered as a formula that is resolved before being used by the rule.   The Apply Adjustment approach enables the analyst to apply the adjustment value to a specific measure variables like ECL. You can also apply the adjustment to specific groups of instruments using attribute-based level filters, such as geography and instrument type.   The measure variable and filter (portfolio slice) for the adjustment drive the combination of aggregation and allocation methods that you choose to apply.     Once you have determined the adjustment type, you will need to apply the adjusted value to your selected measure variable. You can choose to use one of the following methods.         Adjustment Type   Absolute – the absolute value of your measure variable Relative – a % change to be applied Increment – a value to adjust your measure variable   Allocation Method   Even – equally allocated across all records Proportional – spreads based on the proportional value of the measure variable or the weighted variable Individual – applied individually to each record   Aggregation Method   Sum Mean Max Min   Now let's look at an example. In the Allocation Rule Example below we can see a hierarchy starting at Bank 1. Auto Loans is an instrument type with three regions rolling up. Finally we can see how the four Instids roll up to the South Region.     Select any image to see a larger version. Mobile users: To view the images, select the "Full" version at the bottom of the page.   Her the analyst chooses to apply an adjustment for all Auto Loans of 100. Assume that the allocation method is proportional, and the aggregation method is sum.  All adjustments are allocated to the instrument level proportionally and summed up to a new total (inclusive of the other instruments).     The adjustment is applied at Auto Loans. Each child will then get a value of that adjustment based on the proportion of the values. Here we can see that Region has the North with a proportion of 33.3%, South with 41.6%, and West with 25%. And we can also see the proportions of the instrument ID.     The adjustment value is then spread proportionally to all the children of the level where the adjustment was applied. After the adjustment is applied to the Instrument IDs, the allocation with a rule of SUM is applied to the roll up of the values to the parent levels.     So, the South region now has a new sum total of 291.59, instead of the original proportional total of 291.6. And we can continue with that aggregation of Sum to Auto Loans with a total of 699.98. And also, a new total for Bank 1 of 1099.98. Note how the HELOC values did not change as there was no adjustment made to that Instrument Type.     Allocation rules are important to understand to ensure that your adjustments to the Expect Credit Loss is applied to the correct level. And how the allocation type and method will impact the children of the level you selected.   For further information and to view a demo of how to apply an Allocation Rule in SAS Allowance for Credit Loss, please review the training course Using SAS Allowance for Credit Loss.     Find more articles from SAS Global Enablement and Learning here. ... View more

The purpose of this post is to explain how a risk pipeline is used to perform risk analysis. I will identify the sequence of steps to follow to generated the results. We will also examine the risk methods that are available within SAS Risk Engine.   SAS Risk Engine provides a user interface that walks you through the process of performing risk analyses.​   With the user interface, you can complete many of the same activities that can be accomplished through code, including:   defining risk method programs​ creating risk method maps. ​ You use risk pipelines to create the risk environment, score counterparties, evaluate portfolios, and explore results.   Select any image to see a larger version. Mobile users: To view the images, select the "Full" version at the bottom of the page.   You begin with a risk pipeline, and you add functionality to the pipeline by adding nodes to it. Regardless of whether you use SAS Risk Engine through code or the user interface, several input tables are required to perform risk analyses. These tables need to be placed in memory in CAS before they can be used in a pipeline. Data nodes that add the portfolio data and market data to the pipeline are then configured.   The first step, creating a risk environment, is automatically created for you during pipeline execution. There are no actions that you need to take to complete this step.     Methods and maps are objects that you reference in your pipeline. Evaluation and scoring methods are created on the Risk Methods page. Maps are created from the Maps page. In the pipeline. You use action nodes to add these programs to the pipeline.  ​ ​ A method program consists of a set of computational statements that the risk engine uses to compute one or more output variables for specific types of objects in your portfolio data.  Risk method maps show the relationship between risk methods and specific types of objects in your portfolio data.     After you configure a pipeline, all you need to do is submit the run to generate analysis results.  The risk engine performs a few prerequisite tasks. Then it runs all the nodes in the pipeline sequentially.   ​ You can review the results for the pipeline itself and for each node. You can explore tables in SAS Visual Analytics and in SAS Studio. And, depending on the configuration of the pipeline, results might be available for exploration in SAS Risk Explorer.    What are Risk Methods?   So now that we have discussed the Risk Pipeline creation process you can see one key portion of that process is the Risk Method Program. Let's look at the types of risk methods that are currently available from within SAS Risk Engine.   Basically, a risk method is SAS code that consists of a set of computational statements. The code defines one or more output variables for the objects in your portfolio data. A risk method consists of the method code and the objects that the method code references. Objects that the code references include risk data objects, output variables, and models to name a few.    For the method code, you enter SAS code that runs the computations for each instrument of a given type at each market state and horizon. If you plan to evaluate your portfolio or score your counterparties, you must create a risk method or a model group for each instrument type and counterparty type in your portfolio data.   Risk Method Types   Using the SAS Risk Engine user interface, you can create different types of risk methods. There are several more that can be incorporated when using SAS Risk Engine via code.   A computed method creates a single computed value that exists outside of the forecasted time horizons.  An evaluation method runs computations for each instrument of a given type at each market state and horizon. For credit risk, you could use an evaluation method to compute loan losses. For market risk, you could use an evaluation method to price trades. Here’s an example of a simple evaluation method that uses economic risk factors to evaluate and bin loans.    A mitigation method offsets exposures in your instrument data. You associate mitigants of a given type with a mitigation method. PostVar methods calculate additional output variables after the regular variables and the evaluation method's output variables have been aggregated across sub-portfolios. You use a PreQuery method to use output variables from your evaluation methods to calculate additional output variables in the Query Results node. These variables are not stored in the risk environment.  A scoring method runs computations for each counterparty of a given type at each market state and horizon. For credit risk, you might use a scoring method to compute the probability of default for your counterparties.   How are Output Variables used in Risk Pipelines?   Output variables are variables that the risk engine can use in downstream nodes in a risk pipeline and include in the analysis results. Did you know that the method code must explicitly state at least one output variable, with one exception?  For cash flow analyses, you can specify output variables in the method code. You can also specify output variables in the Cash Flow Analysis section in the Evaluate Portfolio node. Or you can specify them in both places. ​   Output variables are used in risk method programs, evaluation model groups, and scoring model groups and in cash flow analyses. For now, we’ll focus on how to configure output variables.     First, you need to identify the output variables in your method code. The next step is to configure each variable according to its type. There are many options available in the user interface to configure your output variables. How you configure an output variable depends on the results that you want to produce. Is the output variable capturing value, exposure, or return among other possibilities?   When you create a new output variable, you first select a risk method type. You can select Evaluation, PostVar, PreQuery, or Scoring as the risk method type.      ​You also need to select a variable type.  You can choose numeric, for which the output variable’s value is a single number.  An array variable’s value is a one-dimensional array of numbers.  If your output variable needs to contain the bucketed values of the cash flow, you would select a cash flow bucket array.  Finally, select the cash flow measure type if the output variable’s value is either the duration or the convexity of the cash flows.  For each variable type, there are optional fields to populate.  ​   ​And there are up to three categories of configuration options available for your output variable. There are evaluation options, aggregation options, and post-aggregation options. So when you create an output variable, you first select the type of variable that it is. Then, when you are ready to configure the output variable, the options that are appropriate for the type are displayed and can be selected.   Suppose you have an output variable for profit and loss. The risk method type is Evaluation. You’d want to configure this variable to be calculated in a way that corresponds to how profit and loss is calculated and represented. ​   For this example, this type of output variable is typically configured so that an existing output variable is referenced as the base variable, the base case value is subtracted, and the tail statistics sign is flipped. So you would select those options. In the user interface, the configuration options for output variables are found under the categories that I just mentioned. Those categories are evaluation, aggregation, and post-aggregation. Each configuration option has a check box that you can select if you want to apply that option. You’ll see that some configuration options are selected by default.    ​The Output Variables page is where you access previously created output variables and where you create new ones. Output variables are global in scope and share the same definition across risk methods and model groups. And the variable’s name in the method code is the same as the output variable’s name.    Risk Engine ships with three output variables: VALUE, EXPOSURE, and PL (profit and loss). SAS has already created and configured these three commonly used output variables to save time and to remove any uncertainty about the way each should be configured. You cannot modify or delete them. ​ ​ The Value output variable is automatically created when you run a pipeline that contains a risk method that sets the _VALUE_ system variable in its method code. But you will still need to add the VALUE output variable to your risk method.  ​ ​Similarly, the Exposure output variable is automatically created for you when you run a pipeline that contains a mitigation map.  Add the Exposure output variable to your risk method.   To Summarize   Your method code must explicitly set at least one output variable with the exception of Cash Flow Analyses, where you can specify output variables in the Cash Flow Analysis section in the Evaluate Portfolio node​ You can select Evaluation, PostVar, PreQuery, or Scoring as the risk method type.   ​ There are many options available in the user interface to configure your output variables. These options vary by type.​ Output variables are global in scope and share the same definition across risk methods and model groups. ​ SAS has created and configured three commonly used output variables for you. ​ Finally, use the Output Variables page to view and manage output variables.      Find more articles from SAS Global Enablement and Learning here. ... View more

The use of parameters with SAS Allowance for Credit Loss can be found at each step of your business workflow. Parameters allow users to adjust an existing script to their own business needs. For example you may be running your credit analysis task and using a probability of default model. Instead of writing a new script for every business need, you can simply add a parameter that would allow the user to adjust the script. A simple example would be adjusting for time; month, quarter, year.   You can define the script parameters using the parameter builder. The parameter builder enables you to choose the components needed for user input within the user interface. You can also arrange the layout of the components, size the components, and define the interactions between the components.​ ​ A parameter component is an input object that allows users to select or define values within a process flow task. For example, when creating a cycle, you can select a calendar date for the start date of that cycle. Modifying the parameters from within the UI can let the users select input values without having to create multiple scripts to account for each parameter component.   Here are some sample parameter builder components that can be used when creating or modifying a script:   Calendar – a calendar widget will pop up and allow the user to select a date. Cirrus Object Selector – allows the user to select a business object type. Dropdown – a list of values that you define will be available to the user in a drop down list. Radio Button Group – provide the users with a selection of objects. Risk Scenarios Table – enables the user to add or remove scenarios from SAS Risk Scenario Manager Dual Selector – user can multi selected from a defined list of items.   On the Parameters tab, the following page sections are available:     Select any image to see a larger version. Mobile users: To view the images, select the "Full" version at the bottom of the page.   Interactions are sets of instructions that can be used to assign variables, apply if/else logic, call and execute functions, and so on. Fundamentally, interactions are programmable scripts that can be created from within the parameter builder. Interactions consist of one or more logical steps.   Interactions can be used to set the values of certain types of components and enable the interaction between different fields based on their entered or selected values.   From the Components pane, drag-and-drop a component into the Builder section. Once the component is available in the Builder section, the component refers to a parameter for your script. Repeat this step to add more parameters to the Builder section.   Create an interaction or modify any existing interactions for the component.   When you are satisfied with your parameters, click Save. After you have created and saved your selected components and interactions, it is best practice to test and validate that the components are performing as designed. You can test and validate the components by creating an analysis run with the script in which you changed the components.   You can then reference the parameters in your script code using the following syntax: ${Params.<parameter ID>}.   For more information on parameters and scripts can be found in the Using SAS Allowance for Credit Loss course.     Find more articles from SAS Global Enablement and Learning here. ... View more

Expected Credit Loss (ECL) is the estimate of the total loss that a financial institution may suffer from loan defaults. In this blog, we will explore ECL and two of its major regulations, CECL and IFRS9. We will also examine how SAS Allowance for Credit Loss can be used for calculating your Expect Credit Loss. Expected credit loss (ECL) is a financial concept that involves estimating the potential losses a financial institution might incur over the life of a loan. This estimation uses a forward-looking approach, considering all relevant factors and information available, such as historical experience, current conditions, and reasonable and supportable forecasts. What is CECL: Current Expected Credit Loss? CECL requires financial institutions to estimate expected credit losses over the life of a loan, using a forward-looking approach that takes into account all relevant factors and information available. This includes historical experience, current conditions, and reasonable and supportable forecasts. The regulation also requires institutions to disclose more information about their credit risk and management practices.   Here are some key provisions of the CECL regulation: Use of Historical Data CECL requires institutions to use historical data, current conditions, and reasonable forecasts to estimate expected credit losses over the life of a loan, providing a more accurate picture of credit risk. Consideration of Creditworthiness CECL requires institutions to consider the borrower's creditworthiness when estimating expected credit losses, ensuring that all factors that may impact the borrower's ability to repay are taken into account. Collateral and Economic Environment CECL requires institutions to consider the asset's collateral and the economic environment when estimating expected credit losses, providing a more holistic view of credit risk. What are the benefits of CECL? CECL provides a more forward-looking approach to credit loss estimation, which will improve the accuracy of financial institutions' loss estimates. It also simplifies the accounting process by eliminating the requirement for incurred loss models and allowing for more timely recognition of losses, simplifying the accounting process. CECL applies to all financial institutions that offer loans, including banks, credit unions, and non-bank lenders. The regulation will have a significant impact on these institutions' financial statements, with some estimates indicating that it could result in an increase of up to 50% in banks' loan loss reserves. What is IFRS9: International Financial Reporting Standard 9? IFRS9 is a regulation that requires financial institutions to estimate their expected credit losses over the life of a loan, similar to CECL. However, IFRS9 requires financial institutions to classify their financial assets into three categories: those that are either held to maturity, those that are available for sale, and those that are held for trading. The regulation also requires institutions to estimate the expected credit losses for each category of assets using a variety of models and methods. What are the benefits of IFRS9? IFRS9 provides a more accurate and transparent approach to credit loss estimation, which will allow financial institutions to make better decisions about their lending practices. It also provides more consistency and comparability in financial reporting across different institutions and countries. IFRS9 is a global regulation that applies to all financial institutions that use International Financial Reporting Standards (IFRS). The regulation will have a significant impact on these institutions' financial statements, with some estimates indicating that it could result in an increase of up to 30% in banks' loan loss reserves. Why Use SAS Allowance for Credit Loss? SAS Allowance for Credit Loss provides a robust solution for credit loss calculation and reporting. With its flexible modeling capabilities and scenario analysis, SAS allows financial institutions to make informed decisions about credit risk and improve their risk management practices. Additionally, SAS provides robust documentation and audit trails for regulatory compliance, reducing the risk of noncompliance penalties. SAS Allowance for Credit Loss makes it easy for financial institutions to create disclosure reports for regulatory compliance. With detailed documentation and explanations of credit loss calculations, SAS ensures that reports are accurate and transparent. Additionally, SAS allows for customizable reporting to fit the needs of each institution, making it an ideal solution for financial institutions of all sizes and types. You can find more information on how to use SAS Allowance for Credit Loss in the Using SAS Allowance for Credit Loss course. ... View more

In this post, we will explore the concept of individual adjustments and their role in credit risk analysis. Individual adjustments allow us to fine-tune the credit risk calculations and adjust them according to specific circumstances.   What are Individual Adjustments?   Individual adjustments are a way to customize credit risk assessments to account for unique circumstances, refining the credit risk analysis and ensuring a more accurate risk assessment.   Individual adjustments are typically applied in a post-processing step, allowing for a more granular analysis and ensuring that unique circumstances are properly accounted for.   Individual adjustments can take many forms, such as industry-specific adjustments for credit risk, collateral adjustments, or adjustments for specific borrower characteristics. These adjustments can help to better approximate the true credit risk and make the analysis more accurate and relevant.   Industry Specific Adjustments   Industry-specific adjustments are used to account for the unique risks associated with different industries. For example, the credit risk associated with the healthcare industry may be different from that of the financial services industry, and industry-specific adjustments can help to account for these differences.   Collateral Adjustments   Collateral adjustments are used to adjust credit risk calculations based on the value and quality of the collateral pledged by the borrower. They help to reflect the credit risk associated with the loan and ensure lenders take the necessary precautions.   High-quality collateral is considered safer and less risky for lenders. If the collateral is of high quality, the credit risk associated with the loan may be lower, and the collateral adjustment can help to reflect this by lowering the credit risk associated with the loan.   Adjustments for Borrower Characteristics   Adjustments can be made for a borrower's credit history. If a borrower has a history of defaulting on loans, the credit risk associated with that borrower may be higher, and an adjustment can be made to account for this.   Adjustments can be made for a borrower's payment behavior. If a borrower has a history of making late payments or missing payments, the credit risk associated with that borrower may be higher, and an adjustment can be made to account for this.   Adjustments can be made for a borrower's financial condition. If a borrower has a low income or high debt-to-income ratio, the credit risk associated with that borrower may be higher, and an adjustment can be made to account for this.   Using Individual Adjustments in the Credit Risk Analysis task within SAS Allowance for Credit Loss   In the Run Credit Risk Analysis task, you can select your individual adjustments data to apply them to the credit risk calculations, refining the analysis and making it more accurate.   Individual adjustments can be used to refine the credit risk analysis and make it more accurate and relevant. They are typically applied in a post-processing step, after the credit risk calculations have been completed.   Individual adjustments can help make credit risk analysis more accurate and relevant. The adjustments can be based on specific credit characteristics, such as collateral or borrower attributes.   Data Requirements for Individual Adjustments   To apply individual adjustments, your data must contain the following columns: Loan Identifier, Facility Identifier, Adjustment Type, Adjustment Value. The SAS Allowance for Credit Loss solution provides sample individual adjustments data that adjusts the expected credit loss lifetime amount. Individual adjustments are a powerful tool that can be used to refine credit risk calculations and account for unique circumstances. By applying individual adjustments, we can better approximate the true credit risk and make more informed lending decisions.   For more information, see SAS Allowance for Credit Loss Sample Data.       Find more articles from SAS Global Enablement and Learning here. ... View more

  Credit loss models are essential tools for evaluating expected credit losses. One such model is the ACL Curves model template, which is based on a curve methodology. This model evaluates credit losses when the Loss Ratio (LR) curve or the Probability of Default (PD) and the Loss-Given-Default (LGD) curves are provided. In addition, prepayment (PP) is also considered through a curve methodology.   Loss Ratio (LR) Curve: The Loss Ratio curve represents the ratio of losses to the total exposure at default. This curve is used to estimate expected credit losses.   Probability of Default (PD): The Probability of Default curve represents the likelihood that a borrower will default on a loan. This curve is used in conjunction with the Loss-Given-Default curve to estimate expected credit losses.   Loss-Given-Default (LGD): The Loss-Given-Default curve represents the percentage of loss that a lender is likely to incur in the event of a default. This curve is used in conjunction with the Probability of Default curve to estimate expected credit losses.   SAS Allowance for Credit Loss uses SAS Risk Engine to execute your credit loss models.    What is SAS Risk Engine?   SAS Risk Engine runs risk calculations and aggregates results over an in-memory grid. What this means is that the software can handle complex calculations very quickly and store large volumes of results.  With SAS Risk Engine, you have the flexibility to configure new instruments, develop new pricing functions, add new risk factors, and create scenarios as needed. Risk Engine lets you conduct complex analyses of very large portfolios. And the results are held in memory so that you can explore them immediately.   SAS Risk Engine gives you a lot of options when it comes to data sources and programming languages. Data sources can be on-premises or in the cloud, such as relational databases and unstructured data. And, you can manage data and create analytical models using a variety of languages including R, Python, Lua, and of course, SAS. As for transparency, the risk environment configuration is transparent. Risk Engine is an open platform accessible via SAS APIs.   Models are executed within Risk Pipelines​ where SAS Risk Engine can take the the model you supplied, your data (portfolio), risk factors, and functions and performs the calculations for your requested output variables like expected credit loss.   The action is done from with SAS Allowance for Credit Loss user interface when you run the Credit Risk Analysis task.   You can add a model through the SAS Risk Engine user interface by building a risk pipeline.   This pipeline will include your model code as well as the following objects if needed:   pre-model code pre-actions code post-actions code post-model code     An example can be seen on the right.       You can also add your model code within SAS Allowance for Credit Loss. You would need to do the following:   create new model provide the model details select SAS as the Engine in the CODE tab, you can enter your SAS model code.   Screen shots can be seen below.     Select any image to see a larger version. Mobile users: To view the images, select the "Full" version at the bottom of the page.   Credit loss models, such as the ACL Curves model template, provide valuable insights into expected credit losses by utilizing curve methodologies. Understanding the Loss Ratio, Probability of Default, and Loss Given Default curves is essential for accurately evaluating credit risk.​ SAS Allowance for Credit Loss allows you options to enter your model code so it can be executed within SAS Risk Engine. If you would like to learn more about the SAS Allowance for Credit Loss or SAS Risk Engine solutions, please use the SAS Risk Learning Subscription to find the associated course material.     Find more articles from SAS Global Enablement and Learning here. ... View more

  The purpose of this blog is to explain the automated demand forecasting used within SAS Intelligent Planning.   I will focus on discussing the demand forecasting process and how forecasting restoration and hybridization will provide you with increased confidence in the demand forecasting process.   The two fundamental messages of how demand forecasting with SAS can help, is: First, lower forecasting efforts leading to cheaper forecasts.  The process tangibly reduces the amount of effort from various departments & people that go into the forecasting process. second, significantly improved forecasting accuracy   How Demand Forecasting Helps   A lot of this improvement in productivity is achieved by SAS’ capability to automate the forecasting process including the ability to select the most suitable forecasting models for every occasion. This allows the user to focus just on the exceptions. The ability to include relevant external data and combine this with the client’s internal data further enhances the improvement in forecast accuracy. Automated demand forecasting of most SKUs will be completed at the relevant item and/or  store level. The solution will automatically test and apply the best fitting analytical models (See 'Model Training using Time Series and Machine Learning Forecasts' section below) for each time series class with model stability. Your various data sources will be integrated to include internal and external predictors. And the solution will use an automatic historical demand restoration (See 'Demand Restoration' section below) when considering product Lifecyle in the forecasting process.     What is the Demand Forecasting Process       Your data is inputted through a preprocessing step that prepares the data for the model forecasting. Preprocessing includes demand restoration, forecast flag calculation and demand driver preprocessing. Model training uses machine learning models and time series models to automatically apply the most accurate forecast. Machine learning models will provide a lower level forecast typically done at a SKU/store level. While time series models will provide a higher level forecast typically at category/region level. The solution provides alert notification and forecast autocorrections as part of the forecast post process. The goal is to ensure low-touch forecasting for the user with embedded forecast autocorrections and prebuilt alerts. Alerts are notifications to check whether the generated statistical forecasts are reasonable or require checking. These alerts can be viewed in prebuilt reports. Auto corrections help detect inadequate forecasts and bring them to a confidence interval. These are very useful when a forecast value is too high or small for a single location/product pair.   How does Demand Restoration Help Prepare Data   During preprocess SAS runs a step called Demand Restoration to recover missing historical values.   These missing data samples are preprocessed as they might decrease forecast accuracy. Demand restoration could be applied to the following examples: Out of stock or insufficient stock periods or periods in the training sample. Promotional periods, considering the degree of the promotion. Phase-out period of life-cycle and history of sales before phase-out. Introduction period of life cycle for regular assortment forecasting. Threshold minimums and various inventory flags help identify stock levels, stock shortages, and promotional periods.   The combination of threshold minimums and inventory flags improves the demand restoration process rather than just using a single out-of-stock flag.   Using those inventory thresholds and flags can improve historical data sales enrichment up to 20%.   Model Training using Time Series and Machine Learning Forecasts   Automatic Time Series Classification will segment the time series models on the bases of patterns that were detected. Each time series model is then classified automatically into one of the pre-defined segments. These segments are based on seasonality, low volume, and intermittent and non intermittent patterns. The benefits of this process is recognizing that different types of times series models require different types of strategies. It improves forecasting accuracy for individual segments or groups within your data. The automated classification of time series will apply the appropriate strategy.   Feature engineering is one of the key tasks involved in the Machine Learning forecasting. These features allows SAS to build learning sets that can incorporate up to 1000 features in the Machine Learning models. Here are a few example of the many features that can be utilized in the modeling process: Promotions – Determines how products behave based on promotions such as price decreases, advertising, or store location for the promo (like an endcap or an aisle display). Calendar Events – such as Christmas, Easter, Valentines Prices – Changes in price over time when the item has not been promoted Lag Features – Looks at how historical shipment volumes at different aggregations impact demand. Analyzing quantities available to be sold compared to demand. Master Data – can use product and location characteristics to view patterns. Here you can see variance is patterns due to regional preferences.     Once the solution calculates the machine learning and time series forecasts, the solution will run the step called forecast Hybridization. This is where the ML & TS forecasts are combined according to defined consolidation logic. We start with the ML and TS forecasts. There are a multiple forecasts created, short term, long term, with and without promo features. Each one as a specific forecast quantity. Using demand classification attributes, the consolidation step creates a forecast mix with the goal of obtaining baseline and total forecasts. The Promo forecast is derived as the different between the total forecast and the baseline forecast.   Summary   In summary automated demand forecasting helps you improve your forecasting accuracy while reducing the forecast related efforts. In the input data preprocessing step, demand restoration can enhance your historical sales data by up to 20%, increasing forecast accuracy. During model training, the baseline and total forecasts are automatically created by consolidating the multitude of created forecast using a process called Forecast Hybridization. The last step of input data preprocessing, contains alert notification and embedded forecast auto corrections that help provide you with a low touch forecast process.   ... View more