Non-maturing deposits (NMD) is a deposit without maturity date. The deposit rate is normally low. Banks could adjust the rate at any time. The customer can withdraw without penalty, however, in real life, the deposit is observed to be "sticky" when LIBOR rate changes.
Now, how shall such NMD amount be modeled for Fund Transfer Pricing (FTP) purpose?
FTP measures the performance of a unit. If the internal funding cost is set as 3 months LIBOR rate, and deposit department could encourage people to put 3 months fixed deposit with (LIBOR rate - 1%) rate, their performance is 1%.
There's a "Stable - Floating parts model" for the NMD deposits: The NMD is divided into 2 parts, a Stable Part, considered as "core balance" and a Floating Part as "non-core balance". The Floating Part is seen as volatile and assumed a very short maturity, e.g. 1 month; while the Stable Part is assigned a longer maturity, e.g. 1 year, and each month, 1/12 of the core part would be considered re-priced.
The assumptions seem too arbitrary to me: why the Floating Part is assumed to be withdrawn in 1 month's time? Why the Stable Part is estimated to run-off in 1 year? And how to decide each part has how big a portion? In real life the NMD amount keep rising, we can't observe a "real" withdraw.
I'm a bit lost now, any ideas please?