The Science of Managing Black Swans

From MIT

If you see thephrase “black swan” and immediately think of either a large bird or thepsychological thriller starring Natalie Portman, you probably are not in thebusiness of managing organizational risk.

To a riskmanager, “black swan” phenomena are highly unlikely events that have massiveimpacts on a business or society on the rare occasions they occur. TheFukushima disaster in March 2011, brought on by a deadly combination of a powerfulearthquake at sea and resultant tsunami, is a sterling example of such aphenomenon. The reactor meltdown that followed was an outcome that some argue could (and should) have been preventedwith better planning.

New researchsuggests that by exploiting many types of data, managers can help prevent (orat least contain) the damage related to black swan events and other risky blindspots. The caveat: organizations should rely less on management experience andintuition and rely more on integrated data to point to potential risks.

How data canhelp risk managers is the topic of a recent paper, Managing Risks with Data, by Ron Kenett, who is avisiting professor at the University of Ljubljana, Slovenia and holds researchpositions at the University of Turin, Italy, and and NYU’s Polytechnic Schoolof Engineering.

Kenett suggeststhat the proper exploitation of organizational data can help prevent some ofthose hugely disruptive, largely unexpected events. In practical terms, thatinvolves acquiring and merging data, as well as building data-driven riskmanagement decision-support systems that complement and reinforce the moretraditional methods used today. According to Kenett:

Risk management is traditionally practicedusing subjective assessments and scenario based impact analysis. This commonapproach is based on experts providing their opinions and is relatively easy toimplement …. Modern evidence based management relies however on data, and notonly opinions, for achieving effectiveness and efficiency. In that context,risk management can exploit information from structured quantitative sources(numerical data) and semantic unstructured sources (e.g. text, voice or videorecordings) for driving risk assessment and risk mitigation strategies.

In his research,Kenett lays out a maturity ladder of risk-management practices:

1.    Intuitive – noformal methods used.

2.   Qualitative – riskassessments are based on expert opinions

3.   Quantitative – somedata is collected and used to derive Key Risk Indicators.

4.   Semantic –unstructured data, like logbooks or blogs reflecting user experience, isanalyzed.

5.   Integrated – datafrom various sources is integrated into a coherent risk management system.

“Manyorganizations are at level 1 or 2,” writes Kenett. “Going up the ladder is botha management and technological challenge.” It’s when organizations are able tocombine the third and fourth rungs — a combo of quantitative and semantic data— to get the final rung of data integration that unexpected risk is bestmanaged.

Kenett, who isalso chairman and chief executive officer of the data analytics firm KPAGroup, suggests that operational risk managementis a function of the complexity of the business and the environment in whichthe business operates. “As a consequence, the more complexity increases, thehigher is the need for integrating internal and external data sources, andfiltering external data according to internal rules and definitions.”

Kennett is notalone in pushing for more data-driven risk management. Bill Pieroni, globalchief operating officer at insurance giant Marsh, contends thatthe best way to manage risk — even black swans — is to use big data.

Pieroni saysthat while they describe once-in-a-lifetime happenings, black swan events —such as the South and Central American defaults, U.S. savings and loan crisis,the October market crash, the U.S. bond market massacre — are happening withgreater frequency than ever. This regularity suggests that some seemingly unknowableevents are, in fact, becoming more-or-less predictable. In other words, blackswan events are “giving way to shades-of-grey swans.”

This is wheredata comes into play. “Analytic competitors who leverage big data willincreasingly be able to identify, model, and act to mitigate or potentiallyexploit these risks,” writes Pieroni.

Even so, Pieronicontends that it’s crucial to distinguish the concepts of risk and uncertainty,two related but distinct ideas. In a 2013 blog, Pieroni wrote that thetwo terms are often used interchangeably, but are actually quite distinct:

Uncertainties pose unknowable and henceunmanageable threats. Risks, however, can be explicitly accepted, avoided, ortransferred. Organizations that are fully exploiting big data are activelyuncovering and converting uncertainty into known risk as well as addressing andexploiting competitive vulnerabilities.

Large, long-livedand historically successful organizations are often most vulnerable toconfusing risk with uncertainty, says Pieroni. One big issue: insularleadership and anecdotal decision-making. “If data and analytics are notexplicitly part of decision-making and outcome feedback, the organization willincreasingly be in jeopardy. Unchanging strategies and tactics work, until theydon’t, with often disastrous outcomes.”