He’s right, though: I can’t believe I didn’t know about it either. That would be Benford’s Law. Wiki summarizes (link):
Benford's law states that the leading digit d (d ∈ {1, …, b − 1} ) in base b (b ≥ 2) occurs with probability proportional to logb(d + 1) − logbd = logb((d + 1)/d). This quantity is exactly the space between d and d + 1 in a log scale .
Less precisely but more intelligibly, in any number of real-life situations, the leading digit will be “one” about a third of the time. Apparently guys who commit financial fraud do not know about Bender’s law: if they invent numbers for (say) a ledger, they tend to randomize, and “one” will lead off only once in nine. So if you see only one in nine, you know somebody is cooking the books. (Hal Varian gets credit for pointing out the fraud application).
I did know a more primitive version: fraudsters also tend to randomize when they are faking social security numbers. But social security numbers are not random: they are assigned by place of birth, so most of my students are in the 900s, and I am 003 (and yes, I know: John McCain’s SSN is #1).
I also remember learning a bit about the base of the natural logs, 2.718281828..., which places a big role in growth-and-decay equations, including those you find in finance. Apparently nobody knows quite where this one comes from, but that it seems to pop up first in financial documents.
The issue came up because my trusted friend Anon said he’d heard that accountants use Fibonacci numbers to track fraud—and then found this. So yes, Fibonacci is involved, but indirectly.
No comments:
Post a Comment