Margaret Fenton,President, Everest Data Research, Inc.
Panelists:
Enid S.Barry, Manager, Technology Evaluation & Integration Services, American Electric Power
Richard L. Fenton, Vice President, Technology Management, Banc One
William M. Hazen, Principal, Bill Hazen & Associates (Management & Information Systems Consultants in the Health Industry)
John H.(Jay) Huey,Senior Vice President, Chief Information Officer, Motorists Insurance Companies
Location:
Motorists Insurance Companies, 471 E. Broad St. Columbus, OH 43215
This month's conversation was hosted by Motorists Insurance Companies, a part of the Motorists Mutual/American Hardware Insurance Group comprising six companies. The Group has assets of more than $800 million. Motorists' personal lines, commercial lines and life insurance products are sold by a network of more than 2,800 agents working through approximately 600 independent agencies in a market of more than 32 million people.
Margaret Fenton, moderator for the meeting, presented two case studies
taken from Computer Ethics by Tom Forester and Perry Morrison (MIT
Press, 1994). The first study demonstrates the need for accuracy of data.
The second study scenario draws attention to the need for the privacy of
accurate data. After each case Fenton asked the panelists and the
audience for comments.
Case 1: Bad Data
In a small town near Paris in 1979, the owner of a service station became
suspicious and notified police of the license number of a car driven by
three young Frenchmen that had a license plate patched together with
pieces of tape. In addition the check they offered had a scrawled signature
on its face.
. . . A routine interrogation of their database revealed to police that the car
had been stolen, and a patrol car was dispatched to intercept them. They
caught up with the young men while their car was stopped at a traffic
light: two officers in plain clothes jumped out, one holding a machine
gun, the other a .357 magnum revolver. The only uniformed officer
remained in the car.
Although the precise sequence of the subsequent events is still not clear,
it is known that the officer with the magnum revolver opened fire on the
trio; the bullet pierced the windshield and hit one of the young men just
under the nose. The other two men were then informed that their
assailants were police (not gangsters), and they were handcuffed while an
ambulance came to assist their injured friend.
Later investigations placed the whole matter in a quite different light. One
of the three men had purchased the car, quite legally, ten days before. It
It was true that the car had once been stolen, but that was in 1976 and it
had been recovered by the insurance company, which had then sold it to
the firm from which the man later legally bought it. The primary cause of
this incident was a failure to update the computer file covering the vehicle
so that changes in status and ownership were accurately represented.
Unfortunately, at the time of interrogation, police records still labeled the
vehicle as stolen, and police reacted as if they were dealing with
potentially dangerous criminals.
Fenton asked two questions,
What could have been done to prevent this problem and
What are your comments about bad data.
RESPONSE FROM PANELISTS J. Huey
I shall respond first since we handle automobile insurance. It is possible
for these circumstances to occur here but it is unlikely. I shall describe
how it might possibly happen. Data is reported to two different places for
a stolen vehicle: to the police and to the appropriate insurance company.
The police report to the National Crime Index (NCI) data base. The
insurance company reports the information to the National Information
Crime Bureau (NICB) data base.
When a car is recovered, the claims adjustor must report to the NICB that
the car is recovered. The police probably know it was recovered because
they
recovered it and the information would go into their file. However, if
someone forgot to record the recovery, then a policeman on duty who
checked the file would believe that the car had not been recovered and the
driver had stolen it.
We do a lot of checking and editing to be sure the insurance file is correct
but the interface between the insurance file and the police data base is not
as good as it should be.
W. Hazen
A next step needs to be taken to assure that data is shared between the
police database and the insurance database.
E. Barry
The more critical the data, the more attention it should get.
R.Fenton
The integrity of a company's data is an issue faced by every company. If
BancOne's data processing was upset for a couple days, it would probably
affect the economy of the nation. There are many processes at BancOne
to insure correctness of data. However it is up to an individual to be sure
his credit rating is accurate since a bank must use this information in
making decisions about granting loans, etc.
The recent death of Princess Diana in an automobile accident has some
messages for us. During the first hour after the accident it was reported
that the car was a stolen Mercedes the had been re-furbished. Shortly the
report was dropped. This was probably due to bad data that came up on
somone's computer screen.
QUESTIONS FROM AUDIENCE What about liability? Who owns the data? How does a victim
identify who is liable? There is the automobile case where an
insurance company has a data file and the police have a data file
about the same car, but the police file is not up to date. Response:
This involves the question of ownership. The insurance company and the
police department each has its own necessary file. The insurance company
makes its information about recovered cars available to the police
department. However, the insurance company can not make the police
update the police file. It would seem to make more sense for a police
officer out on the road checking a car to be able to call into both files.
What about companies that make a practice of adding incorrect
charges to an individual's credit card bill that do not belong to that
individual? Response:
The bank with that person's credit account accepts the charge from an
outside company as a legitimate charge and pays the company. The
charge is then sent to the individual by the bank. When the individual
calls in with a complaint, the bank has ways of checking on this and
tracking down such companies. There is also a center that will assist with
this problem. When a bank finds a company doing this, it will not deal
with that company anymore. It is important for individuals to check their
credit card bills for items that should not be there and to report incorrect
charges to the bank. Don't pay the bill automatically without checking the
entries. The bank should provide help to the individual who has these
problems.
Case 2: Protection of Privacy: Indivdual Rights vs. Company Needs
and Society as a Whole
Scenario:
. . . In 2005 the government has begun conducting compulsory AIDS
testing programs to help predict the course of the epidemic and to
determine the impact of various initiatives. Test results are said to be
totally confidential, and in order to prevent avoidance of the test, Social
Security numbers, tax returns, and databases of all kinds are used to track
down individuals and to present them (forcibly, if necessary) for testing.
Of course, as a law-abiding citizen you present yourself, you are diagnosed
as antibody positive, and the outcome is logged onto a government
database system. However, you are confident that because the outcome
of the test is confidential, you will be able to lead a fairly normal life, you
follow safe sex practices, and hope that in the time you have left a cure or
at least better treatments will be found.
With the growth of AIDS as one of the more common terminal diseases,
however, an increasing number of people have been lying to insurance
companies about their infection, and upon reaching a terminal stage, they
kill themselves in car accidents or other apparently non-suicidal forms of
ending it all. In this way, victims' families are securing large insurance
payouts, and the insurance companies are feeling the pinch. However, in
the interest of individuals' rights, the government prohibits insurance
companies from demanding AIDS tests on an insurance applicant. It is
sufficient that government knows who is infected, and the public release
of such information is considered socially destabilizing.
Nevertheless, insurance companies are desperate to discover such
information so that they can offer competitive premiums without such
high degree of risk. Two options are open to them: to somehow gain
access to government files or else to obtain the same information through
alternative sources. Like insurance companies everywhere, the U.S.
companies hate risk, so they choose to do both. Through the tried and
trusted methods of bribery through a third party, some of this information
is obtained. The remainder is gained by indirect means. Medical
insurance companies note the kinds of treatments that patients receive
and, by looking for patterns in these ailments, quickly determine the
chances that particular policy holders and applicants might have AIDS.
This information is sold to other insurance companies as well.
Furthermore, in order to secure lower premiums for their businesses,
employers are encouraged to monitor workers' use of sick leave, any
gossip about sexual preferences, and any drug usage. These reports, too,
are added to the information mosaic gathered by the insurance companies.
Lastly, landlords and other employers also are eager to obtain access to
these records. In the case of landlords, they don't want to provide
accommodation to an AIDS victim because, given the hysteria over AIDS,
they may not be able to rent their premises again. In the case of
employers, they don't want to invest heavily in training and providing a
career structure for an individual if that person is likely to die before the
investment is recouped. Furthermore, being an AIDS victim makes a
person susceptible to blackmail, and that risk cannot be tolerated for
employees entrusted with heavy re- sponsibilities and financial powers.
COMMENTS FROM PANELISTS
In the hypothetical scenario the government prohibits insurance
companies from demanding a test for AIDS from insurance
applicants (But
no such prohibition exists in fact.). At present our life insurance
company does require an HIV test, a blood test for people applying
for insurance greater than $100,000 in face amount. By government
regulation we must get a signed statement from the applicant saying
it is all right for us to do this. If we get the information back: yes,
this person is HIV positive, we tell the person the insurance is being
denied based on something in the bloodwork without being specific
about it. We refer them to the lab that did the test and suggest they
be re-tested, or challenge the lab, or find out what the lab found out.
We also report this to a collection agency for insurance companies so
that another company will see there is a code indicating that there is
something wrong with the blood work. It does not say this person
has HIV.
Access to many company databases has many layers of
protectionÑnot just one firewall. Employees are screened and are
restricted to the area of their work. Of course there is the hacker
who has all kinds of time to work at breaking into files.
If you think your health record is confidential today, think
again, certainly not in a hospital or a doctor's office.
SOME QUESTIONS AND ANSWERS What about data mining of personal information by consulting firms
and others?
Answer: Data mining can put together trends and place you in those
trends based on information about you. This information can be obtained
in many ways and can be used to try to sell products to you. It also has
useful purposes in helping to focus medical developments by knowing
how many have arthritis, AIDS. etc.
Do we need a way to monitor the information in files about
individuals? Answer: There are so many files, there is not going to be a way to
monitor them all. Indeed, many more files will be developed.
Do we need the consent of an individual to use data that exist in
files? Answer: In the event of a medical emergency it may not be possible to
get the consent of an individual and have it forwarded to an appropriate
source that has the appropriate medical file and perhaps other medical
sources in order to obtain the medical data needed to save the life of the
individual.
Comment: Are we over-obsessed with the privacy of data?
Council for Ethics in Economics
125 East Broad Street
Columbus, Ohio 43215-3605 U.S.A.
(614) 221-8661 FAX: (614) 221-8707
