can the firms’ product managers judge the likely success rates of
different products? Which kinds of products (and marketing campaigns)
have been most successful in the past? Based on surveys and market tests,
how should the companies reassess their products’ chances of success?- In the relentless pursuit of quality, a parts supplier for the automobile
industry seeks to reduce its rate of product defects. How does it
estimate its defect rate? How can it identify the key factors that affect
this rate? Would modifying its production-line process reduce the rate? - Do a chemical company’s emissions into the air (at levels within legal
standards) pose a health risk for its workers or the surrounding
residents? Are they responsible for an increased rate of certain types
of cancer in the community? Is the cancer rate actually elevated, and,
if so, what other factors (age or other characteristics of the
population, or even chance) would account for this?
In the preceding examples and in most other similar problems, there is
no shortage of historical data that may have a bearing on the probability being
estimated. The tough questions are: What is the best way to interpret the data?
How can the manager identify factors that distinguish when a risk will be high
or low? These are not easy questions to answer. Nonetheless, the road to the
answers almost always begins with constructing two-dimensional tables of prob-
abilities. Such tables look much like those in the wildcatter example. The col-
umn headings list the actual risk or uncertain event of concern to the decision
maker. The row headings summarize the way in which the decision maker has
categorized the data—identifying factors that influence the relevant risk. The
next section presents a typical example.Predicting Credit Risks
How can a bank assess and accurately predict the credit worthiness of a new
business customer? In recent years, banks have increasingly turned to statisti-
cal measures, compiling computerized composite credit “scores” for customers.
The bank’s aim is to distinguish high- and low-risk accounts, closing or reduc-
ing credit limits on the former and increasing limits on the latter.
Consider how the method works in screening traditional business loans.
The loan division of a bank has spent considerable time and energy devel-
oping a scoring system for predicting the default rates on different loan
accounts.^3 The scoring formula incorporates key characteristics of the cus-552 Chapter 13 The Value of Information(^3) For an account of banks’ scoring methods, see M. Quint, “Banks Raise Scrutiny of Credit Cards,”
The New York Times,March 27, 1991, p. D1, and R. Simon, “Looking to Improve Your Credit Score?
Fair Isaac Can Help,” The Wall Street Journal, March 19, 2002, p. A1.
c13TheValueofInformation.qxd 9/26/11 11:02 AM Page 552