I have seen a lot of posts on how to create a financial model. But how do you check a model done by somebody else?
This is how I do it:
(1) Validate the change in the forecast with simple math
No matter how complicated the model, somebody will try to do math in their head. This can help you determine if the model is wrong but does not mean that the model is correct necessarily. If simple math cannot be applied, explain why.
(2) Examine the historical baseline
Although this seems obvious, I have seen a lot of models which either don’t have enough data, the data is not reliable, or the trend is unforecastable. However, the last thing you want to say after you see actuals is, “we should have known this based on the historical data.”
(3) Examine the logic
There is no easy or quick way to do this, but I like to follow the math through to the lowest level assumption to make sure that its correct, and the logic is intuitive. I used Control-[ in Excel a lot.
(4) Understand the variables that are not captured
Most models require simplification but it’s helpful to know which factors are not explicitly included and how these factors are accounted for in the model implicitly, or not.
(5) Understand the story behind the numbers
Is the story consistent with the story told to investors and the Board? If not, decide if the story or the model is wrong.
(6) Change the assumptions and examine the output
Some models break when the assumptions are changed either because they only work in a narrow range, are highly sensitive to changes or due to model error. This is one of the easiest ways to determine if the model is broken.
Leave a comment