Recent cognitive science has taught us that an overreliance
on unchecked intuitions and professional judgment can lead to persistently
suboptimal decision processes and even inefficient markets.
price complex risks, adjust claims, investigate premium leakage
and claims fraud, make hiring decisions, and guide risk management activities. As in Moneyball, inefficient domains that
traditionally had been both data rich and analysis poor continually are being made more efficient by using data analysis
tools as correctives for systematically biased expert judgment.
While the business analytics revolution is popularly attributed to “big data” and the exponential growth of computing
power, Kahneman’s book teaches us that in many quarters, the
rise of business analytics is also attributable to the shortcomings
of System 1-style thinking. Unlike human experts, computer algorithms can objectively weigh multiple sources of information,
and do so without getting tired or being fooled by random noise,
predictively irrelevant details, or deceptive narratives.
Indeed this has been a theme of psychological research for
more than 50 years. In the 1950s, Paul Meehl pioneered the
study of clinical vs. statistical prediction, repeatedly finding that
even rudimentary regression equations can perform the unaided
judgment of physicians—and by extension baseball scouts and
insurance underwriters. (See “Analyzing Analytics: The Debate
Between Intuition and Institutional Thinking,” July/August
2008 Contingencies, for more details.) Kahneman, whose subsequent work sheds considerable light on Meehl’s findings, writes
that Meehl was one of his heroes when he was a young student.
All of this has an important implication for the actuarial
profession. Building models to help experts—in many domains—make better decisions is a very natural extension of
actuarial practice. (In fact, the title of a classic paper that Meehl
co-authored is “Clinical versus Actuarial Judgment.”) An appreciation of Kahneman’s themes in the age of big data and
cheap computing power presents the actuarial profession with
an opportunity to considerably expand its footprint.
The Illusion of Validity
Of course all of this can be difficult in practice. As Lewis il-
lustrated in Moneyball, it takes time and effort to change an
organization’s culture to favor data-driven over intuition-driven
decision-making. Some of the reasons for this are straightfor-
ward. It is difficult to learn statistics and perhaps even more
difficult to do statistics well. Statisticians and business people
often speak different languages. And many organizations en-
counter difficulty attracting and retaining people with the
suitable skills. Creating data marts and putting knowledgeable
people in charge of data stewardship similarly require invest-
ments in time and money.
