Driver Frequency Rules
This article explains driver frequency rules in Model Reef and how driver frequencies interact with model periodicity and variable timing.
You will learn:
How driver frequency is defined.
How drivers are aligned with the model timeline.
How upsampling and downsampling of drivers works.
Good practice when mixing different frequencies.
Driver frequency is important for getting sensible behaviour when you apply drivers to variables over the life of the model.
What is driver frequency
Every driver has a frequency, for example:
Daily.
Weekly.
Monthly.
Quarterly.
Yearly.
The frequency tells Model Reef how often values are expected and how they should be aligned with the model's base periodicity.
Driver frequency is independent of model periodicity, although they interact.
Model periodicity vs driver frequency
The model has a base periodicity, such as monthly or weekly. Drivers may be at:
The same frequency as the model.
A higher frequency (for example daily drivers in a monthly model).
A lower frequency (for example annual drivers in a monthly model).
Model Reef handles alignment as follows:
If driver frequency equals model periodicity, values map one to one.
If driver frequency is higher, values are aggregated into model periods.
If driver frequency is lower, values are spread or held constant across model periods, depending on the driver type.
This happens automatically when drivers are used in variables or formulas.
Downsampling drivers
When a driver has a higher frequency than the model, for example daily prices in a monthly model, Model Reef will:
Aggregate values into the model period using an appropriate method.
Common patterns include average, sum or period end, depending on the driver type and how it is used.
Examples:
FX or price series used as rates are typically averaged over the period.
Volume or count series may be summed.
The exact aggregation approach depends on how you design the driver and formulas.
Upsampling drivers
When a driver has a lower frequency than the model, for example annual inflation in a monthly model, Model Reef will:
Spread or step the values across the higher frequency periods.
Hold annual values constant within the year or interpolate as appropriate for the driver.
Examples:
Annual inflation rate applied monthly as a constant rate per month.
Annual growth index mapped into a smooth monthly trend.
In many cases, it is clearer to define drivers directly at the same frequency as the model to avoid confusion.
Choosing driver frequencies
Guidelines:
If you are forecasting at a monthly level, prefer monthly drivers for most assumptions.
Use annual drivers only for very high level long term assumptions such as terminal growth.
Use daily or weekly drivers only when the pattern within the month is important for the decision you are modelling.
Using too many mixed frequencies can make it harder to understand the shape of series.
Changing model periodicity and driver impact
If you change the model's periodicity, for example from monthly to quarterly:
The model timeline is regenerated.
Drivers are reinterpreted at the new period boundaries.
Variables recompute using whatever driver frequencies exist in the Data Library.
Because drivers are stored independently of the model timeline, you can change model periodicity without reimporting them, but you should always review key driver behaviour after such changes.
Related articles
Last updated