I have wondered this myself. A friend of mine (former senior 747 pilot for United) told me when I asked that he could think up two reasons: 1) cost, and 2) FAA foot dragging.
I have used transponders in research for decades. They are somewhat more expensive, but they will last for years (we have had them last for 3-4 years, depending on the amount of use.
Also 37.5 kHz (which I believe is the pinger frequency) is fairly short range in the ocean. Normally, frequencies used in deep-ocean work are 12-13 kHz or less.
Of course, there is this question: Is the Malaysian airline paying for the search?/s
You forgot cost per benefit.
Almost never does this scenario happen where no one knows the location of the crash within days. The 30-day minimum is in itself a fudge factor. I’d put money down you can’t find a single other crash that wasn’t found within the requirement period.
So, why waste money on a scenario that has a .1% chance of happening? (This conservative site should understand that, as we are supposed to oppose the “if it saves just one life!” safety-NAZI drivel.)