I would guess that it was a marketing decision. Running the bus at 32 MHz might have increased costs in two ways. The motherboard would have to work at the faster speed and a handful of chips would have to work at the faster speed.
Manufacturing costs for the parts would have been pretty much identical--except that yield might have been a bit lower if they disposed of the ones that wouldn't operate at 32 MHz.
Back then, it might have had a noticeable impact on yield which would raise the cost by some percentage, but it might not have. While I don't know where the technology was at that time, I can guess based on what was available on the market. The IIci had been around a couple of years running just fine at 25 MHz. I doubt that the state of the art was having any trouble with 32 MHz by that time.
However, I really am not sure how much of a challenge 32 MHz vs. 16 MHz operation was back in that time frame. Today, the price would be identical. The tools are available to design and build hardware for operation at either speed with trivial effort.