Nassim Taleb has been telling people that banks tend to sit on time bombs while convincing themselves they are conservative and nonvolatile.

True: documented public statements by Taleb dated before 2008-04-03 (books such as Fooled by Randomness or The Black Swan, articles, interviews, talks) repeatedly warn that banks or financial institutions carry hidden risks of rare large losses (blow-ups) while appearing stable or low-risk. Mostly true: such warnings exist before that date but are fewer, or are aimed at financial institutions or risk models generally rather than at banks specifically. Mixed: there is only one or two passing remarks of that kind. Mostly false: his earlier writing concerns tail risk in general but does not apply it to banks seeming conservative. False: no such statements before April 2008 can be found in his published works or interviews, as established by a review of those works.

Set before the research began.

1. E1 188-189_BKR_0907 The Banker · September 2007 Own record

   “The problem is that the second type of risk, which we can call blow-up risk, is far more vicious, mainly because of its sinister, hidden nature.”

2. E2 188-189_BKR_0907 The Banker · September 2007 Own record

   “Clearly the problem can be generalised to governance for banks and public companies: think of two of the primary banking problems in recent US history – the 1982-84 debt crisis and the savings and loans crises of the late 1980s.”

3. E3 [PDF] Mathematics of Black Swans House Science Committee (testimony by Nassim N. Taleb) Own record

   “Banks are now more vulnerable to the Black Swan than ever before with “scientists” among their staff”

4. E4 Month: February 2009 artdiamondblog.com · February 2009

   “He warned that while the growth of giant banks gives the appearance of stability, in reality, it raises the risk of a systemic collapse — “when one fails, they all fail.””

The only evidence dated before April 2008 is one September 2007 article in The Banker (E1, E2), which describes hidden blow-up risk and applies the problem to bank governance. E3 is undated and E4 does not date the warning it reports, so the evidence cannot show whether such warnings were repeated, fewer, or only one or two remarks, as the criteria require.