Here are two reasons:
1. Noise models not correct (probably not the case).
2. The bias circuit noise is irrelevant for the output of interest. Could be like common-mode signal for a differential output. To further test this hypothesis, run an AC sim with the input being the bias current and the output as your output of interest.
Depends on your bias circuit model. An ideal current source is noise free, you have to use real transistor models to model current source noise. Apart from this issue, the noise contribution from bias circuit in a good designed opamp should be small (but non-zero).
If the circuit is symmetric to the biasing there is no noise from the bias generating circuit. That could lead to crazy noise effects. For instance if you have a differential signal the noise amplitude will be modulated with the signal. That is likely if you have a bias current generating source which has, mostly because of supply current efficiency reasons, a much smaller current. The noise is current is proportional instead of square-root.