Creating a risk model

You can create a risk model to help predict the possibility and level of impact of potentially risky events on the financial outcomes for an entity, individual or business.

About this task

Each entity, whether individual or business, represents a certain level of risk depending on where they come from, what work they do, what type of transactions they perform, and so on. When an institution on-boards an entity, or an entity comes to the institution asking for a particular financial product, the institution needs to know what kind of risk they carry. For example, if a business comes to a bank asking for a business loan, and the bank finds out that the entity does business in a politically unstable country, the entity poses a greater risk. Based on several risk indicators, a risk model can help the bank make an informed decision about whether to approve the loan, or whether to on-board a potentially risky entity.

Procedure

To create a data model:

  1. Log in to the IBM FCII user interface and then select Design Studio.
    Note: Users assigned to Administrator and Data Scientist roles have authorization to access the IBM FCII Design Studio.
  2. On the Risk Model tab, click Create Model.
    Tip: If a model exists similar to the model that you want to create, you can view the model and then click Create New Version to update it.
  3. In the form that is displayed, enter the name, description, and lookup code. Select decision and operations services from the drop-down lists, and then select a model type (Bayesian Network or Rule).
  4. Click Create to create the model. After a model is created, you can click the Show/Hide pane in the lower right of the page to display existing risk indicators. For more information, see Working with risk indicators.

Example

For basic risk scoring based on client type, products/services type, and geography type, the IBM FCII Design Studio provides a Promontory Risk Model, which categorizes entities into Low, Moderate and High risk types. This model considers risks for an entity to come from the following risk categories:
  1. Client Risk has two risk factors, called indicators: Client Type, and Industry (or Occupation) Type
  2. Products and Services Risk has the risk indicator: Product (or Service) Type
  3. Geographic Risk has two risk indicators: Associated Countries and Out-of-Country Address
The risk indicators are assigned risk weights as per their relevance in the total risk a client carries. Each risk factor also has risk scores assigned from 1 to 3 (1 being Low risk and 3 being High risk).

A weighted average of all these factors gives the total CRR for an entity, the score falls into one of the three categories Low, Moderate or High based on the ranges mentioned in the CRR legend.

After a user inputs the information about an entity pertaining to the risk indicators, the RMS engine returns the required CRR for the entity. Each call to the RMS engine is made through REST API and each call is asynchronous. In brief, RMS does not maintain context of individual calls and does not store risk score histories.