This is purely anecdotal, based on personal experience:
When the local Verizon network is congested the download speeds can drop precipitously while the upload bandwidth is still very good for the peplink equipment. At the same time, a mobile device (we deploy Apple products) will see an equal or better download bandwidth, with a far worse upload bandwidth.
If we add up- and download bandwidths in these situations the experience is that the peplink router combined bandwidth is significantly better than the combined personal device bandwidth.
I would conjecture that this reflects Verizon traffic shaping, prioritizing download capacity for personal devices at the cost of their upload capability, while routers are free to upload as fast as they can, though still limited their in their download bandwidth by the overall allocation wthin the download bandwidth available.
As I said: Conjectures based on anecdotal experience… YMMV.
[Update: The account of the router SIM cards rolled over just now, and the download bandwidth improved 5-10-fold. Thus the conjecture that in our case the cause was traffic shaping by Verizon (deprioritized after the monthly threshold had been exceeded) seems plausible.
Two things of note: 1) the upload bandwidth seemed unaffected by the deprioritization, and 2) the deprioritization had a far more severe consequence for our location than what we had anticipated. It was not occasional, but rather close to constantly in effect for the remainder of the billing cycle.]