Let's say that you want a mortgage on a house that has not been recently bought, there is no definite market value to it, so the bank uses a model based on comparable houses recently sold. That is one of the simplest needs for models.
But models are also needed for risk analysis. A publicly traded stock is one of the most liquid, price discovered things on earth. Giving a loan on its current value can be a huge mistake. If the stock suddenly drops, the lender is left with an asset that cannot possibly cover the loan. A lender would use a model to see how much the asset can reasonably lose in value, and how likely it is to lose value along with other assets.
Mark to model can be used to find fraud in the market. If
@Earl and
@Stone sell back and forth a piece of land inflating its value, the model can show the true value of the land.
What you will notice about all these uses of models is that they only reduce the assumed value of an asset, never increase it. The market should be treated as a high point, with the model being a way to reduce that number.
A free market cannot exist as free if the participants cannot disagree with it. All buyers and sellers have a model in their head for the price, and if the market does not meet or exceed that price, they will not buy or sell. This allows the market to take a poll of the group wisdom.
