Booz Allen is now saying that the merger failure rate is lower than the two-thirds number trumpeted by Booz (and others) in years past. One of things they point to is the increase in consolidating deals, which - at least in theory - should be likelier to succeed than deals involving new lines of business.
These conclusions are anecdotal, although we may see some data to back it up in the next year or so. Still, seems like a lot of the consolidating deals I've seen (and the one I lived through) in the last couple of years have been at very full prices.
If by "failure" you mean selling the acquired company for a fraction of the purchase price several years later, I agree that you won't see much of that from consolidating acquirors. However, if you define "failure" as a merger failing to meet the IRR assumptions that led to its approval, the high prices paid of late lead me to believe the numbers really haven't changed much.