I've spent a lot of time recently investigating on Benford's law. Many sources state that Benford's law can be useful when auditing accounting data. Non-fraudulent accounting data (under "good" terms) should follow Benford's law. Now my big question is: Why is it so??? What I'm looking for is a mathematical model in business/accounting that explains the phenomenon (unlike the usual roughly/heuristic approaches). No matter what, I really did not find one (searching the internet). This resource https://en.wikipedia.org/wiki/Benford%27s_law for example states some reasons for Benford's law to appear:
- Outcomes of exponential growth processes
- Multiplicative fluctuations
- Scale invariance
- Multiple probability distributions
I suppose that "Multiplicative fluctuations" could be the reason why accounting data conforms to Benford's law. But I'm not happy with this vague answer. I am missing a mathematical model explaining that behaviour. "Mathematical model" means: random variables, assumptions about them, their interaction.
Is it clear now?