Skip to Main Content
Return to Navigation

Understanding Stratification

Stratification enables you to summarize large volumes of financial instruments to a manageable scale for processing and reporting purposes. The engine that does the processing is the Stratification application engine (FI_STRATIFY). This application engine enables you to perform mathematically intensive calculations on a relatively small number of instrument pools that are proxies for a much larger number of individual instrument records. Using this feature is a multistep process, which includes:

This section discusses:

Group by Operations

When creating pools of instruments, you need to identify the operation for each of the attributes. You have the following choices in terms of how to populate each attribute in the instrument pool from the instrument data:

  • Discrete: Directs the Stratification Engine to populate the instrument pool with discrete values.

    This option populates the pool with discrete values for an attribute if there is no logical or mathematical way to group instruments with different values, and the attribute is significant for reporting purposes. Choosing the discrete action for an attribute ensures that only instruments with matching values are grouped together in a pool. The discrete action may be the correct choice for important code fields. Be careful when using the Discrete user on amount, date, or rate fields. These types of fields tend to have many discrete values. A typical value for a discrete group by is an ID field or a yes/no flag.

  • Default: Use for attributes that you want to set to a hard-coded value.

    If you are defining a stratification rule that you know is used only by one class or type of product, and you want all pools that are generated from that rule to have the same value for a specific attribute, select this option to force the attribute to have that predefined value. The Default option should be used with caution and only when you are sure that the value is valid for all pools that are created by the stratification rule. When you are unsure, a better choice is to either drop the attribute entirely, or populate the field with discrete values.

  • Increment: If you want to group the possible range of values into consistent ranges, the simplest grouping operation to use is increments.

    Identify the increment value, and Stratification application engine that groups the instruments into incremental strata. For example, you can define the interest rate field to stratify by using increments of 0.25 percent. If all of the instruments have interest rates falling into the range of five to six percent, Stratification Engine creates four strata or groupings of interest rates, 5.00 to 5.25, 5.25 to 5.50, 5.50 to 5.75, and 5.75 to 6.00. You can use the Increment user on numeric fields, date fields, or both. For example, you can stratify the start date, end date, first payment date, and first repricing date.

Note: The Stratification application engine is preprogrammed so that it always processes the following attributes by using the discrete action: product ID, price index ID, variable rate, accrual basis, compounding frequency, payment frequency, and all unit of measure fields. Be careful to select the discrete action for required attributes. This is because the Stratification application engine must create a pool for each unique value that it finds in the data, and for every attribute which is chosen to use the discrete action. Examples of attributes that probably should never be assigned the discrete action are: balance amounts, start or end dates, payment amounts, and customer ID.

Action Operations

These are the default action operations:

  • Average: Calculates the average value of an attribute for all instruments in the pool.

  • Weighted Average: Calculates the average weighted value for an attribute by the current balance.

    Examples of attributes for which you may want to calculate the weighted average are interest rate fields: interest rate, cap rate, floor rate, and interest rate margin for floating rates. Other possibilities include remaining term, original term, loan-to-value, and so forth.

  • Min, Max (minimum, maximum): Select one of these operators to populate the pool with the minimum or maximum value of a given attribute for all the instruments that are part of the pool.

  • Mid: Calculates the Mid Date value for all date values in the pool.

    [max (date) ? min (date) / 2 ] = Mid Date

  • Sum: The sum action user calculates the total sum value for a numeric attribute.

    The most common attributes that you want to calculate the sum for are the balance and payment amount fields. By calculating the total of the balance and payment amounts, the total of all pool balances should reconcile to and represent the totals of all instruments that are in the portfolio.

Stratification Tips

Here are some criteria to keep in mind when evaluating how you want to stratify data:

  • When identifying attributes that you want to include as part of the instrument pools, keep in mind the primary use of the stratification rule that you are creating.

    For example, Cash Flow Generator uses only the base currency equivalent balance and amount fields for its calculations. As far as Cash Flow Generator is concerned, designing a stratification rule that populates other balance or amount fields is optional.

  • You need to identify the stratification criteria for the attributes.

    This step is simplified by the fact that any attributes that you identify as needing to maintain discrete values for, or that are going to be dropped, or defaulted by definition, cannot be used as grouping criteria for pools.

  • Maximize efficient processing by designing rules to summarize the instruments into as few pools as possible, while at the same time only grouping instruments that generate the same or a similar stream of cash flows.

    In other words, all of the instruments that are grouped together into a pool should share common and pertinent financial characteristics.