By Simon Firth, Nov. 2005
Experts say it threatens the CIA, the US military, corporations
and governments around the world.
In a recent speech to the Joint Services Conference on
Professional Ethics, Lt. Colonel William Bell of the US
Army called it "possibly the greatest danger facing
us today. It undermines everything that our Army is supposed
to value and stand for."
What is this menace? It’s the fear of taking risks.
It’s not something you'd expect to find in the gung-ho
military or in supposedly buccaneering industries like
high tech. But the problem is very real in both, says Bernardo
Huberman, Senior HP fellow and director of the Information
Dynamics Lab at HP Labs.
Whenever an organization gets to be larger than just a
handful of employees, Huberman explains, its executives
often become risk averse – even when the riskier
of two possible actions could considerably benefit their
"Most organizations consider taking risks as an essential
part of their success," he notes. "Yet few have
policies to encourage risk-taking by managers and employees."
During the past year Huberman and Labs colleague Tad Hogg
have studied this problem and developed a novel idea for
solving it – what they're calling "decision
An offshoot of their lab's ongoing research into market
mechanisms for pricing utility computing and the dynamics
of electronic communities, decision insurance could be
a model for how to get executives to do what corporations
want and need them to do – make smarter decisions
about the risks they are asked to take.
It's not hard to see how the problem of risk-averse decision-making
In the military, says Huberman, "if you're concerned
about promotions in your career, the important thing is
not to have a single blemish on your record."
Similarly, he says, sales managers are often faced with
the choice of pursuing an old account or spending time
on a risky, but potentially profitable, new line of business.
This happens in part because individual employees' interests
may not always be aligned with those of their organization,
adds researcher Hogg.
"If you're an individual salesperson," he suggests, "your
bonus usually depends on whether your particular sales
happen or not. But from the company’s perspective,
one particular sale is not as important as the total sales
for the quarter or the year."
In such cases, explains Hogg, an employee is being rewarded
for an outcome (closing a particular sale) and not for
the decision behind it (which, if different, might have
opened up a new market for the company).
One way to make people focus on decisions rather than
outcomes is to allow them to feel free to fail – in
other words, to take more risks.
"It’s not that we want to make people more
risk seeking," cautions Huberman. "It's that
we want them to be risk neutral. Then they will be able
to evaluate potential opportunities in a rational and objective
But how should you do that? Well, why not offer insurance
against making a wrong (i.e. unprofitable) decision?
In a proposal outlined in a recent
paper Huberman and Hogg recommend
grouping decision-making executives into a pool, just
as insurance companies pool drivers or
Each group member would pay a regular premium into the
insurance pool, which would be used to compensate anyone
whose bonus, perhaps, was lessened because of decisions
that didn't pay off.
"This shifts the risk from you to the organization," Huberman
says. "It encourages you to be less risk averse, and
it aligns your interests with those of the company."
There's an obvious problem with this idea, though. If
I have such insurance, what's to stop me from making a
series of not just risky but bad decisions? However much
I blundered, I'd still be covered, right?
This is a common issue with any insurance – known
as the problem of Moral Hazard. How Huberman and Hogg solve
this problem for decision insurance is one of the most
innovative things about their idea.
With house or auto insurance, laws prevent Moral Hazard
from being a problem. It's illegal to burn down your house,
for example, and then claim the insurance.
In the case of decision insurance, says Huberman, the
answer is to tap the peer network of an insured employee.
Every member of an organization has a group of people, "who
create a consensus as to whether or not you’re doing
your job," he says. "This is why people worry
about their reputations.”
These people, along with your manager, are in a best position
to know how serious you are about your work and what your
track record is, says Huberman.
To make sure the group is a true reflection of an employee's
peers and not just a selection of their friends (or twisted
by another party to include only their enemies) Huberman
and Hogg tapped into some of their lab’s work on
the dynamics of electronic communities.
In a widely publicized experiment, they found that by
simply analyzing the ‘To’ and ‘From’ headings
of internal e-mails, it's possible to establish the group
of people with whom anyone most often interacts.
Often, that group has very little to do with a person's
organizational position. "You might belong to Lab
X, but have most of your peers in Lab Y," says Huberman. "And
your manager might not even know that."
An established peer group can offer insight into the quality
of a member's decisions, independent of whether those decisions
paid off in terms of outcome (e.g., money made, military
battles won, legislation passed or customers served).
By doing this you are now rewarding effort rather than
outcomes, adds Hogg.
"The easiest thing to compensate for is outcomes," he
says. "But what an organization would really like
to figure out is, are you making good decisions based on
the information you have? A mechanism like this can encourage
people to put the company's interests first when they’re
making their decisions."
In some ways this model is like profit sharing, where
some part of all employees' compensation is based upon
the company's overall profits in a given period.
In very large organizations, the connection between what
any individual does and the company's profit is often extremely
"With the decision insurance," explains Hogg, "you
can choose the appropriate group size. So among a group
of sales people, for example, any individual can have a
big impact on that group –and thus on their bonus
as well. This way, when you are deciding between a more
or a less risky project, you understand what the result
will be for either your insurance payout or profit-sharing
Hogg imagines that such insurance might be a mandatory
part of a particular group's compensation package, or it
could be something that people have the option to buy – it
would depend on how risk averse they were as a group in
the first place.
In the latter case though, what if the only people who
take the insurance are the 'losers' in a group? This is
another well-known problem in traditional insurance, called
Although the example used in the researchers' paper assumes
that all individuals are the same, Hogg says he expects
that the insurance mechanism would evolve so that premiums
could be adjusted based in individuals' track records.
In extreme cases, when people are really bad at their job,
they might not be offered insurance at all.
Given the proprietary information that decision-insurance
adjusters would need to set realistic rates, it's most
likely that such mechanisms could only be set up within
But that also makes this kind of insurance a realistic
Hogg and Huberman were inspired to think along these lines
by the work of influential Yale economist Robert Shiller,
who has argued that even in economies such as America's
and Europe's, people are generally underinsured. Shiller
would like to see people offered career insurance, for
The idea is exciting interest. Huberman recently presented
the group's work to the business school at the Federal
Institute of Technology in Zurich, Switzerland, and at
an international conference on risk in Toledo, Spain.
“Many of the executives that come to these seminars
complain about the same problem," Huberman reports. "Their
subordinates are too risk averse, especially in Europe."
Will HP ever adopt such insurance? Huberman and Hogg aren't
"The role of research is to create options for the
company," notes Huberman, "and this, certainly,
is an option."
Simon Firth is a writer and television producer living
in Silicon Valley.