Updated: Oct 22
Why am I sharing a picture of a model airplane if I am talking about financial models? I can try to make an analogy - both require planning, following a process, attention to detail. To be honest, though, when I think of building models, I remember my father, an aviation enthusiast who really enjoyed building model airplanes.
I never had the patience to build toy models but, oddly, I can sit for hours at a computer creating financial models. I've worked on quite a few financial models over the years and in this series, I will share what I've learned about pro forma financial models. To begin with, some general comments about financial models.
WHAT IS A MODEL?
I found some good - although perhaps a bit dry - definitions of qualitative and quantitative models in a blog post titled, appropriately "Qualitative and Quantitative Modelling" on the KBManage website.
"Models are tentative pictures of relationships that attempt to describe causality at points in time and space (Byrne, 2002)"
"Quantitative models are compact representations where a single differential or difference equation may describe the performance of the system for a large set of input functions and initial states (Lunze, 1998)."
"Qualitative models do not require mathematical formulism, but are used to “draw, diagram or represent visually ideas, hunches, perceived patterns or relationships between parts of their projects, discoveries in their data…and so on” (Richards, 1999)."
So basically we use models for two reasons:
To understand how something works.
To understand how different parts of the model interact with each other.
A COLLABORATIVE PROCESS
An effective model can only result from a collaborative effort between the modeler and the user(s). In other blog posts, I wrote about the risks of errors in developing spreadsheets.
Keep in mind that a user who understands his business well can immediately tell if a spreadsheet is wrong. He does not need any knowledge of spreadsheet structure or formulas. He knows his business well enough to spot an error in a calculation or in how two data points are related in the model.
In such cases, it does not matter if the developer has not made any mistakes in creating the model (which is not likely) - the model is still wrong.
NO BLACK BOX
An effective model does not use a "black box" where important information is hidden from the user. For example, using a macro to perform a calculation, then placing the result in a cell, leaving the user to wonder what the number means or how it was calculated.
Users must trust that the numbers in a model are correct. This means they must be able to see and understand the assumptions that form the basis of a model. A black box model makes that impossible.
What are the assumptions, relationships, and numbers on which the model is based?
How does each of these inputs affect the results?
What is the relationship among these various inputs - how does a change in one cause a change in the others?
It should be easy to modify when the types of decisions change.
For example, initially using units per hour to measure productivity, then adding or changing to units per hour per employee as the productivity measure.
The model is used daily, monthly, and yearly by managers to make critical decisions:
Who to hire or fire
Which projects to fund
Which products to manufacture
Whether to sell, buy or lease capital equipment
Over time the decision-making process will change and improve. The models must be flexible so they can evolve and remain an effective tool in making decisions.