Why Does the AM AM Plot of a 3rd Order Product of AMP_B Not Have a Slope of 3 as Expected

Why does the AM/AM plot of a 3rd order product of AMP_B not have a slope of 3 as expected?

This typically occurs when IP3 is much more than 10 dB above P1dB. What is happening is the amplifier has a fairly small linear region, ending far below P1dB. This is because of the effects of the 3rd and 5th order terms of the polynomial, which are relatively large in order to model the specified IP3 and P1dB.