two's complement question

I'm learning about binary and two's complement.

I cannot replicate the result from the following problem with my current knowledge of two's complement - subtraction:

*"subtracting -5 from 15 is really adding 5 to 15, but this is hidden by the two's complement representation:"*

Code:

` 11110 000 (borrow)`

0000 1111 (15)

− 1111 1011 (−5)

===========

0001 0100 (20)

The question is: in the last 4 bits (left nibble), how can you borrow from 0 if it's 0? The example above seems to be borrowing from the next colum where the top bits are all 0's. Can someone explain how this is happening?