Good question, since, in principle the shear lag problem is relevant both for compression and tension. Perhaps, the commentary to AISC-360 give an explanation why the treatment is different for the two actions.

The introduction point of a load (or a moment) in some body can bring high stress. In the case of bolts, near to the connection holes,  traces of plasticity and sagging can be found. But this makes the plate thickness grows, as it is pressured near the bolt surface (contact zone), suppose that the plate don't tear due to high local pression. This deformation can improove the resistance a little. But this happens in tension as in compression. However, when in tension, the end part of the channel, for example, can be forced to cut out from the body. In compression, this will not happen, because you have all the body to resist the force. Also, when in compression, due to global and local buckling, the forces will never reach yielding stress. In tension the design allows to reach  0,9 fy stress which summed to the existing residual stress can lead to yielding. So I believe that shear lag is a load introduction fenomena and happens in both cases. In compression, the local and global buckling occur at lower loads, first then any damage caused by shear lag. That's why it is not a matter of concern when designing for compression, in general.

Tensile stresses can  cause fracture.  Compressive stresses do not unless they cause shear in a connection(think block shear failure in bolted connections). I will get back to shear later.  Design of tensile members members are therefore governed not by yielding but by the net section fracture at their connections. Net section fracture capacity will be very low if only a part of the cross-section transfers the tensile load. Shear lag has this effect: it denies  contribution of a large part of the cross-section and instead concentrates high stresses on only a small part of the cross-section. Obviously, net fracture capacity, which depends on net area available,  is compromised.  Now why "shear"?. Draw the stress trajectories, you will see that the uniform tensile stresses change to shear stresses  at the edges of the hole for example. Shear stresses  move in and can be  quite large in magnitude. In bridges, their magnitude can easily develop cracks . Compressive stresses,  by the way, tend to  close cracks.  Designing tensile members for yielding of the member itself, is a dangerous practice. 

Good question.. Please share me the best answer might you trust...

Regards…