Vintages - Time x Time

One of the more unique conceptual challenges when designing an Essbase or Planning model can revolve around handling multiple Time-basis dimensions. Many model requirements need granularity beyond the standard observation date that is typically defined as being the Time Period in a model. Consider the following:
A customer has a portfolio of investments, each having several characteristics related to some measurement of time.

• The origination date of the investment
• The maturity date of the investment
• An expected length of the investment, based from which the origination or maturation date can be interpreted
• An activity date

A single Time dimension (or two time dimensions that split portions of time apart, i.e. separating Years from Months and Quarters) can not accurately capture metrics in the previous example like the performance over time of all investments that originated in a given period, or how much return at a certain age of an investment can be expected.

Multiple Time Dimensions
Assuming the standard Essbase or Planning time dimensions represent the observation date, additional time-style dimensions can be created to provide more granularity. In the previous example, a dimension indicating the age of an investment can be utilized to analyse performance of investments based on how long ago they were originated. Likewise, identification of the starting or ending dates of investments allows analysis based on similar origination or maturation dates.

How much is too much granularity?
That all depends on the analysis requirements for a model. Adding more detail to a model will increase its size and complexity, but will provide a more detailed picture of time's effect on activity. All decisions regarding expanding a model require careful considerations of the benefits of additional complexity.



Other Time Time appears multiple Times (Time beyond core modelling requirements)

• Scenario revisions

First pass, second pass, etc: These are essentially observations or snapshots of a process at a point in time. Depending on how many of these snapshots are required, it may be advantageous to have a separate dimension identifying when the snapshot was presented. Consider the benefits carefully of disposing of forecast revisions. Often, one key long term goal to improve forecast reliability is the ability to track variances between revisions over time.