This article should be subtitled "How to lie with statistics". The author is comparing absolute numbers of bank failures over time, and completely ignoring the fact that not all bank failures have the same impact. In particular, if a failed bank has 100 times as many customers, its failure will affect 100 times as many people -- and the failing banks of 2008/09 are much bigger than the failing thrifts of the 80s and 90s.