Page 80 - BFSI CHRONICLE 3092022.indd

BFSI Chronicle, 11 Edition September2022
th

statistical probability of hits received on the various selection bias. It can lead to overly optimistic beliefs
parts of the aircraft body (fuselage, tail, engine, wings because multiple failures are ignored. As Taleb puts
etc.) and its chances of survivability even with being it- The highest performing realization will be the
hit, strengthen that part so that even after being hit the most visible. Why? Because the losers don’t show up!
plane survives. It could have looked something like
the inset picture and the obvious conclusion would If we look around there are plethora of examples
have been- shield the wings and tails. which fall into the trap of this bias. In 2001 Jim Collins
came out with a book “Good to Great: Why Some
Companies Make the Leap...And Others Don’t” - has
been an oft-quoted book in non-fiction, management
and finance genre. Jim Collins culled 11 companies
out of 1400+ companies based on the parameter of
their stock beating the market over a 40-year time
span. He then gleaned common characteristics among
them that he believed made for their success. In
another book “In Search for Excellence” Tom Peters
and Robert Waterman, suggest some 8 common fea-
tures of 40 odd companies. Both of these are suffering
from the same ailment- survivorship bias.

Again, why? You may ask.
However, Wald had a counter intuitive argument-
the planes being observed are the ones that have Since, both lack objectivity; rather than first deciding
survived; the ones that got hit elsewhere—eg. the on the parameters/ features/ characteristics and then
engine—didn’t get back and therefore were not searching for the companies to justify the claims, the
included in the U.S. military’s analyses. Thus, there’s books take the other way round- they first search for
no data on the planes that didn’t survive. If the planes the companies and then state the common features-
returned even with a hit on that part (see pic), then effectively a post hoc analysis. In fact, research found
that part is already strong and thus might not need that through 2012 the stock of six of Collins’s 11
the reinforcement. Long story short- Wald’s deeply “great” companies did worse than the overall stock
mathematically & statistically significant argument market and of companies culled out by Waterman &
towards this was accepted, and as the cliché goes- the Peters, about half of the publicly traded stocks did
rest is history. worse than market in coming periods. Both, thus
highlight, that the system of analysis was fundamen-
Survivorship bias- the selective distortion of tally flawed(Shermer, 2014).
truth
What’s stated above is one of the most famous This sort of bias is very much evident in the stories of
accounts of WW II- shrouded in legend and excellent successful leaders, motivational speakers, sportsper-
storytelling (Casselman, 2016). It involves counter son and first-generation entrepreneurs who started
intuitive thinking, mathematics and notions of con- their businesses from their garages. No one writes
ditional probability. Nevertheless, it triggers the books on the failed entrepreneurs who dropped
basic definition of “Survivorship Bias” (sometimes out of college to start their own business. In most
also referred to as Survivor Bias) – a logical error of cases, the successful entrepreneurs create post hoc
concentrating on visible things lying in front of us narratives that explain how they turned their dreams
and overlooking or ignoring those that did not, typ- into reality, evading the part luck, pure chance and
ically because of their lack of visibility. It is a form of circumstances played in their success. And, because

The Institute Of Cost Accountants Of India

80