Have you ever thought the following when looking at a model?
- I am not entirely sure what this model is trying to do, but I don’t like to ask as I am sure everyone else understands it.
- This long formula looks impressive and I don’t want to show my ignorance by asking what it is doing or if it could be made shorter.
- I really don’t understand what all these buttons are on this model, but I expect they are calculating something clever. I’ll press them and hope for the best.
- I am not really sure why it takes so long for this model to produce results for different sensitivities. But I expect that there is a reason.
- I am prepared to make decisions on the basis of this model as I believe the results coming out of it. I assume that the modeller knows what they are doing.
I suspect that the above thoughts have often crossed the minds of those who use and write financial models.
However, please consider the following:
- In almost every exercise on my financial modelling courses it is perfectly normal for all the delegates to get different answers. They are all intelligent and motivated people, but there are so many mistakes that they can make, even in a training environment. Most tell me that they are shocked at just how many errors models can contain and that they will probably never trust a model again.
- In many professional exams, the pass mark is 50%. If a financial model is 50% wrong, then there is a serious problem.
- How many modellers and bosses are blissfully unaware of modelling subtleties? How about the knowledge of different calculation methods? What about the timing of inflation and interest? And what about the difference between NPV and XNPV (and no, it’s not just a matter of XNPV being more accurate because it can introduce specific dates).
How can models be improved?
- Never make the assumption that Excel is an easy tool to use.
- Make sure that financial modellers are properly trained. “On-the-job” or sub-standard training is a very risky approach.
- Only employ financial modellers who are logical, pay attention to detail and have good memories.
- Make sure that error checking is systematically built into models.
- Challenge at all times.
Above all, never make assumptions. And remember George Box’s wise words:
“All models are wrong, but some are useful”