HP 12c Platinum Error 0 in Percent Difference

[Kaijikunito]

http://h10032.www1.hp.com/ctg/Manual/bpia5184
2005 User Manual
Appendix B: Algebraic Mode (AlG) p. 241

I recently acquired an HP 12c Platinum calculator (I miss the gold plate!) and I am doing a read-through of the manual in order to understand how to use it.

On page 241 of the manual, the following instructions are given:

"To find the percent difference between two numbers:
1. Key in the base number.
2. Press [-] to separate the other number from the base number.
3. Key in the other number.
4. Press [change %]

Example: Yesterday your stock fell from 35.5 to 31.25 per share. What is the
percent change?

Keystrokes
(ALG mode) Display
[CLX] [CLX] 0.00 Clears any pending operations.
35.5 - 35.50 Keys in the base number and separates it from the other
number.
31.25 31.25 Keys in the other number.
[change %] -11.97 Nearly a 12% decrease."

I perform [f] [REG] before attempting these calculations. When I follow the above instructions, I receive an Error 0 display, which occurs when y = 0.
Note: In RPN, this yields -188.03. ***

I also receive the Error 0 display when I follow the above instructions with an added [ENTER] after 31.25. Thus:
35.5 - 31.25 ENTER change% = Error 0
Note: Occasionally when I perform these keystrokes, -88.03 is displayed
but I cannot consistently replicate this result. ***
Note: in RPN, this yields 0.

I would like to note that when I perform the following two sets of keystrokes in Algebraic mode, I receive the correct answer. This leads me to question whether [-] can always be substituted for [ENTER], as stated in the manual:

35.5 ENTER 31.25 - change% = -11.97
Note: These keystrokes yield an Error 0 display in RPN. ***

35.5 ENTER 31.25 change% = -11.97
Note: These keystrokes yield the same result in RPN.

So my question is, am I doing something wrong or misunderstanding the instructions?

Although I understand that RPN mode requires different keystrokes than ALG mode,
I am curious as to where the answers -188.03 , -88.03 and Error 0 come from in my RPN notes (marked with ***) so that I can better understand the stack.