What to consider when valuing a company
Have you ever wondered why analysts from different broking firms will value the same business quite differently? It often comes down to the assumptions they make, and the level of simplicity or complexity they build into their models.
When valuing a company, it is important to find a balance between simplicity and complexity. A model that’s too simple may be easy to read, but it could miss key turning points. In contrast, one that is too complex could be less accurate because it’s built on too many assumptions. As investors, it is important to find an appropriate level of detail to maximise accuracy.
One way of simplifying a model would be to aggregate products to a group number. This way you can study revenue and margins on a group level. This leaves you with two simple drivers to predict and understand; eg. for a retailer this would be volume of sales and average margin per unit sold. For a company selling largely homogenous products this could be effective. Particularly if the products have similar underlying drivers. Eg. Demand for tables and chairs are both driven by consumer discretionary spending. In this scenario it may make sense to look at group metrics such as sales per square metre.
However, when products are more heterogenous this assumption can be dangerous and miss critical inflexion points. Taking an average margin assumption won’t be accurate if there are significant differences amongst products and the mix shift changes. This can occur if there are different drivers for the products; eg. jet fuel vs car fuel. Let’s assume jet fuel margins are greater than those of car fuels. Now if jet fuel volumes grow faster than car fuels, your average group assumption would understate earnings growth. In contrast, if lower margin car fuel volumes grow faster, you will be overestimating earnings. Over-simplifying could lead to missing a product mix shift. This is particularly important for products with different supply and demand drivers and different margins.
A simple solution to predicting this more accurately would be to model volume and margins for both jet fuels and car fuels. But why stop there? Is there a difference in margins for commercial vs retail fuels? How about which state they are sold in? If we estimate all of these aspects our number of assumptions (or degrees of freedom) increases significantly. We’ve gone from predicting 2 drivers (volume and margins) to 64 (2 initial drivers x 8 states and territories x 2 fuel types x 2 consumer groups). With each additional layer of complexity you are forced to make more assumptions. Each of these assumptions adds variability to your forecasts and having too many can reduce your overall accuracy.
As investors it’s important to find a balance between keeping things simple, yet having enough complexity to spot inflection points.