10 Rational Risk Measures
A model risk is anything that can potentially cause a model to produce erroneous results. Below are 10 ways to measure model risk.
Risk metrics help us make rational decisions regarding names, formula length, helper columns, and more. I use an XL app to extract these measures from various models to determine which method is least risky. If you would like this app connect with me on LinkedIn and ask for it. I’ll reply with the link.
Are these your top 10? Post you thoughts on the Beyond Excel Facebook page.
Changes – Each change adds risk. Modeling creates risk. Measure: Count changed cells.
Formula Proliferation – Every identical formula behaves exactly the same way; thus, we only need test each distinct formula. Measure: Count distinct formulas
Complexity – Each formula function and operator adds complexity. Measure: Count distinct formula functions and operators
Transparency – Each step required to find a formula’s source values and meaning decreases transparency. Measure: Count steps to precedents
Consistency – We expect formulas in row (traditional calculations) or column (tables) to be identical. Auditors shouldn’t assume this, but modelers shouldn’t make it easy for auditors to make mistakes. Measure: Count inconsistent formulas
Structure – Calculations flowing in normal reading style are easier to audit. Measure: Count out of order precedents.
Labels – All numeric values should be labeled so we can understand their meaning. Measure: Count missing labels.
Data validation – Data validation helps prevent erroneous inputs and associated erroneous results. Measure: Count input cells missing validation
Mixing Inputs with Calculations – Keeping users out of calculations helps prevent inadvertent changes. Measure: Count mixed worksheets.
Unprotected Calculations – Locking calculations helps prevent inadvertent changes. Measure: Count unprotected calculation worksheets.