ELC: liberty file output capacitance index different from simulation

When generating a liberty dotlib for a bi-directional buffer, I have noticed that the cap index differs from what was used in the simulation.  This only affects the PAD of the buffer and not the internal output.  It appears to add the capacitance of the ouput pin(PAD) to the simulated value.

 Using this in my setup file:

Index    CLASS_A  {

        load =  3.0000p   4.1667p   5.3333p   6.5000p   7.6667p   8.8333p   10.0000p  ;

 I get the following from the generated liberty file.  Not capacitance of pin is listed in the dotlib as 2.124p

        index_2 ("5.12392, 6.29062, 7.45722, 8.62392, 9.79062, 10.9572, 12.1239");    p

 When the bi-buf is simulated as an input, the core output pin cap index match my internal index load, also noted that core output pin cap is 0.

Is there a way to turn this off to not add the total net capacitance of the pin and index load.  I'm trying to compare modifications to an IO cell and see if they affect timing, however because the PAD ouput has changed slightly I get different indexes.

Thanks.