Models in Accounting

February 9, 2009 by · 1 Comment
Filed under: Video 

Is it dangerous to use models in accounting? Douglas Wilding discusses this and the risks associated with models and estimates in accounting.


One Response to “Models in Accounting”
  1. Frank Tilghman says:

    There are several issues with models.
    One, models may be developed for one purpose, and then be adapted for a similar purpose. An example of this is credit scoring. This model was originally developed as an indicator of probablility of a borrower filing bankruptcy. The lower the score, the more likely the borrower would file bankruptcy. The model was then adapted to determine creditworthiness, the higher the score, the better the credit. In this case, the model works well for its original purpose, and not so well in its adapted purpose.
    Two, models depend on quatifiable data. What happens when the data is not quantifiable, or ambiguous, or depends on human behavior? For example, many prepayment models predict that for a certain lowering of interest rates, mortgage refinancing activity will accelerate, consequently increasing prepayment speeds. But what if consumers are scared? Even though the model predicts they will refinance, the model can not predict that the consumers are reacting to other factors.
    Models are useful in telling what has happened, but less useful in predicting what will happen.

Speak Your Mind

Tell us what you're thinking...