When to build a model
People build models to help them understand big structures when all they have is a possibly incomplete collection of parts. Like an architect’s scale model. You can’t live in it, but you can find that there is not enough light in one corner, or that it seems not elegant, or that it falls down.
The model provides insight into how the parts may fit together, work together, or could work together better. Often some pieces show themselves to be unnecessary. Sometimes a substitute must be found because of the price.
Do models work?
It depends on what you mean by work.
If you mean finding insight, yes, they usually work. If you mean do they predict the future, no, they don’t.
The common error is to expect too much
No model includes all possible conditions and all models rely on conditions continuing as they are. You can never create a complete environment and test all possibilities within that. The future will be different.
Models cannot contain all possible realities.
A model which took account of all the variation of reality would be of no more use than a map at the scale of one to one. – Joan Robinson
We all build informal models
At least mentally. You have a budget for this month even if you have never written it down. You know the effect of an unforeseen expense. If you have a more formal budget you may know the effect over several months. A budget is a simple cash flow model.
Part of our problem with models arose with Aristotle
Aristotle claims a thing must be true or not true. There is no third condition. The “excluded middle” has meaning in ancient logic, but little application in real life. If a model can be shown to be not right, does that automatically mean it is wrong? Aristotle says yes, but reality says there may still be useful insights to gain.
We know all models are wrong to some extent. How wrong must one be before it is useless?
Recall the purpose.
If a model uses twisted raw data, we can likely assume it is wrong. It is very difficult to draw a meaningful inference from data that does not include historic reality. If days the stock market is up are included, but days that are not are set to zero, we would not likely want to base our investment decisions on the resulting model. No insight to be drawn here.
If a model intentionally excludes data that would contradict the intended premise it is wrong. If we survey welfare recipients and ask about the appropriate level of benefit, we will get a different answer from when we survey all adults. Some insight, but the data is too biased to rely on the modeled answer.
Any model that is created to prove some point is wrong. Models are for insight alone. They point to things to be further understood and tested. No model proves anything.
Models are part of the risk management process.
The risk definition from yesterday matters.
“Risk means more things can happen than will happen”
Elroy Dimson, London Business School
A well-considered and well-designed model will help you see more of the things that are possible. Perhaps change your idea of probability of occurrence. You make better decisions when you can estimate more of the possibilities.
Be careful of detail. Details obscure more than they illuminate.
Don Shaughnessy arranges life insurance for people who understand the value of a life insured estate. He can be reached at The Protectors Group, a large insurance, employee benefits, and investment agency in Peterborough, Ontario. In previous careers, he has been a partner in a large international public accounting firm, CEO of a software start-up, a partner in an energy management system importer, and briefly in the restaurant business.
Please be in touch if I can help you. firstname.lastname@example.org 866-285-7772