"brett" wrote:
If I divide two negative numeric cells and put the
result into a percentage cell, it will positive, even
if the change is negative. For example:
A B
1 -6249 -5810
A1/B1 = 107.56%
The change is moving from B to A. That means I
should have a negative change.

What makes you think so? -5810*107.56% is -6249,
the correct answer. If you used -107.56% arbitrarily,
you would get +6249, which is wrong.

How can I show that?

Since your choice of negative "change" (factor)
seems arbitrary, I don't know what you would
want in all cases -- for example, B is 2 and A is
-4, and B is -2 and A is 4. Note that when B is
4 and A is 2, the "change" (factor) should not
be negative.

Your request makes more sense to me when we
are talking about actual change, not growth factor.
Whenever A is less than B, we might reasonably
want to express the percentage change as negative.
But we must use that notion of change carefully.
For example:

To compute percentage change (C1):

=IF(B1=0, A1, SIGN(A1-B1)*ABS((A1-B1)/B1))

To apply (use) percentage change to B1 (D1),
which should equal A1:

=IF(B1=0, C1, B1+SIGN(C1)*ABS(C1*B1))

Examples (B1=before, A1=after):

A1 B1 C1 D1 (should = A1)
2 4 -50% 2
4 2 100% 4
-2 -4 50% -2
-4 -2 -100% -4
2 -4 150% 2
-4 2 -300% -4

Some people will quibble with my choice when B1=0.
It is arbitrary.