How many failures are we talking about? And how disproportionate are the failures to the percentage of micros you've installed?
My experience does not corroborate any problems with micros. But then, I don't really have enough examples to make a comparison. I can only think of 3 electrical panel failures in the 7 years that I would have known about such a thing in our company. They were all with micros, but the majority of our installs at that time were micros so that essentially means zilch. We had one job where a half dozen panels started showing signs of cell cracking, but the distinguishing feature in that job was the brand of panel. (It was with micros; the panels showed signs of lowered output but none of them outright failed).
Regarding your current case, if the panel voltage is ~2/3 of what it should be, you have a failed internal string of cells. (It's not bad bypass diodes. The diodes are doing their job or the entire panel would fail.) Of the three failures I mentioned above, 2 were this.
One possibility is that you have the same failures with string inverters but you just have no way of noticing them. If you have panel level monitoring with micros and a panel's output drops by a third or two thirds, you notice, or your customer does. If you have a 13 panel string and the voltage drops by one 39th, then nobody notices, or can readily prove that it isn't just dirt or the weather or panel degradation.