Hi Pritesh.

When dealing with NMP systems (let us assume continuous-time systems), the problems you may encounter are due to zeros characterized by positive real part.

In this case you have to deal with the following problems:

 x) inverse response;

 x) reduced phase margin that implies limited bandwidth and slower closed-loop response (a positive zero is somehow similar to a delay);

As long as the phase angle is concerned, it largely depends on the other singularities of your transfer function.

Regarding the disturbance rejection capabilities, it depends on wether you consider low-freq. disturbances or high-freq. disturbances. Since your bandwidth will be limited you should not have problems with high-freq. disturbances. On the other hand you may have problems rejecting low-freq. disturbances, but this largely depends on the low-freq. shape of the modulus Bode diagram of your trasfer function. More in depth, the higher the modulus is in the disturbances freq., the better the rejection will be.    

I agree with Matteo. Another issue not in your list is robustness when the system is moreover unstable: if an unstable zero is not very far from an unstable pole, the controller will have a poor robustness whatever the control design.

For non-minimum phase systems, it is called unstable zero. This type of systems is deemed difficult to control. These systems are often associated with an inverse response at a step change of the input. It is possible to stabilize, to replace the term "integral" by a term "derivative" and use the Routh criterion or the Nyquist diagram for the stability margin analysis.