You're asumming the final number id managed by a sole number. Maybe there a "size" value and a "quantity" value.
An example would be a float representing the final magnitude over the base magnitude (i.e. thousands of dollars), and the base quantity of thosands of dollars an unsinged 16bit integer.
It would be a commonly considered weird approach (you could/should use Decimal types for money), but hey, the final number is strangely near to be such a typical number (2^16).
PS: Like a sibling suggested, other hypothesis is that maybe it was something akin to a bare napkin calculation/decision made by humans (with human errors) eventually digitalized.