DEFEDNING AGAINST BUSINESS CRISES WITH THE HELP OF
INTELLIGENT AGENT BASED EARLY WARNING SOLUTIONS
Shuhua Liu
Institute for Advanced Management Systems Research, Åbo Akademi University
Lemminkäisenkatu 14 B, DataCity 6th floor, FIN-20520 Turku, Finland
Keywords: Crisis Management, Early Warning Systems, Business Life Cycle, Intelligent Agents, Intelligent Data
Anal
ysis Methods
Abstract: In the practice of business management, there is a pressing need for good information management
instruments that can constantly acquire, monitor and analyze the early warning signals of business crises,
thus effectively support decision makers in the early detection of crisis situations. With the development of
advanced computing methods and information technology, there bring new opportunities for the
construction of such instruments. In this paper, we proposed the use of business life cycle model as a larger
framework of guidance for an early warning system of business crises. We also developed a framework for
an intelligent agent based early warning system.
1 INTRODUCTION
“…It is not a question of if or whether an
organization will experience a crisis; it is only a
matter of what types of crisis will occur, what form it
will take, and how and when it will happen” (Mitroff
et al, 1996).
In the age of globalisation, information explosion,
fast-paced technology development and market
changes, uncertainty abounds. As a result, the
possibility grows considerably for various crises to
emerge. Thus, crisis management is becoming an
increasingly important item on the agenda of many
organizations, public and private, large and small.
Crisis is an event or a situation that has considerably
negative affects on an organization as a whole. A
crisis may have the potential to even destroy an
entire organization. To be more specific, in this
paper we shall focus on business crisis that happen
in the corporate world. By business crises we refer
to the highly abnormal, risky, and difficult situations
(e.g. severely shortage of key resources - human,
money, material) that are threatening or could
threaten to seriously interrupt or harm business, and
to bring tremendous damage to the company, its
employees, its product, service, assets, as well as its
reputation (cf. Barton; Bernstein Crisis
Management). Examples of the events that can put a
company into different types of crisis situations
include: environment pollution, product defects,
inappropriate merger and acquisitions, senior
management staff moving to competitors, break-
down of information system that result in halt in
business operations, legal cases, fraud that ruins
business reputation.
Crisis management usually encompasses a chain of
activities that enable a business to plan for, to
respond to, and to recover from a crisis event, so that
eventually it can avoid a crisis to happen, or reduce
the damages as much as possible, or even turn crises
into opportunities (Mitroff et al, 1996; Augustine,
2000; Gao and Yuan, 2003). These activities can be
generalized as three important parts: pre-crisis, in-
crisis and post-crisis activities. Crisis early warning
is a pre-crisis activity. It aims for the early detection
and alerting about highly risky and harmful future
events, so as to prevent crisis to happen as early as
possible. It is also the focus of this study.
Crises like less harmful risks are a natural
companion of any business development. They are
sometimes simply inevitable because the
development of anything is never linear, there are
always ups and downs. Crisis creates complicated
management situations, and they will become more
and more complicated if not controlled at an early
stage (Mitroff et al, 1996; Gao and Yuan, 2003). So
58
Liu S. (2005).
DEFEDNING AGAINST BUSINESS CRISES WITH THE HELP OF INTELLIGENT AGENT BASED EARLY WARNING SOLUTIONS.
In Proceedings of the Seventh International Conference on Enterprise Information Systems, pages 58-65
DOI: 10.5220/0002543600580065
Copyright
c
 SciTePress

the very nature of crises makes an early warning
system that is constantly functioning much
important and desirable. An early warning system
will form the very first defense against any
foreseeable crises and it implies a proactive
approach to crisis management. Without an early
warning system, even if a company is already in
crisis, the management team may still be unaware of
it. Thus, sufficient ability in the early detection and
warning of various types of crises is a must for the
healthy and sustainable development of any
business.
In this paper, we will elaborate on how can IT and
IS help in improving an organization’s early warning
capability, and how intelligent computing methods
and intelligent agent technology can be applied to
build crisis early warning solution that will better
manage the early warning tasks. We shall first draw
upon existing theory and empirical knowledge on
business crises and early warning signals, and then
discuss the relevant methods and models for early
warning analysis. Based on that, we will examine
the role of soft computing methods and intelligent
agent technology in the construction of computer
based early warning support instruments for constant
acquiring, monitoring and analyzing of the early
warning signals, and further the prediction of
impending crisis.
2 EARLY WARNING OF
BUSINESS CRISES
Business crises can be originated from inside a
company or outside the company. Crises situations
caused by sudden changes and devastating events
outside the company, for example in the political
and economic environment, the natural environment
and the social environment, are uncontrollable by
the company. But crises caused by particularly
harmful events happened inside a company are often
attributed to mistakes and failure of the management
structure and management function (Mitroff et al,
1996; Gao and Yuan, 2003).
Although the importance of early warning may seem
too obvious, in reality it is usually a step skipped
altogether in the corporate world (Augustine, 2000).
There may be many reasons behind it, for example, a
top management mind that believes that they are
somehow luckier than others, or that they can handle
whatever crises that may come. Other than that, the
inherent challenge in establishing an appropriately
large early warning system is definitely also an
important contributing factor. Firstly, the number
and form of the crises that an organization may
encounter are practically unlimited. In this sense an
early warning system can never be perfect because it
is impossible for any such system to not overlook
every possibility, even with the best resources.
Secondly, organizations are constantly bombarded
with signals of all kinds. This implies that the efforts
of probing and scrutinizing into business operations
and its management structure must be made
constantly instead of only a one-time effort in order
to capture early warning information. And, the
various signals need to be accurately assessed to
discern the critical ones from the disguising ones.
Thirdly, crisis warning signals may come from
different sources and they may present conflicting
contradictory evidence. How to integrate various
information, to understand the connection between
the seemingly isolated information, and to correctly
understand the “situation” thus to recognize whether
there is a crisis or not, are even more challenging
tasks. Fourthly, warning signs of crises are often
dispersed throughout the organization as well as in
the business environment. However, to be effective,
early warning efforts as well as other crisis
management activities need to be systematic.
“…Effective crisis management is not a function of
how well early warning is done in one part of an
organization, or the sum of separated parts, but a
coordinated integrated whole… In a crisis, poor
performance in one area is not compensated by
exceptional performance in another.” (Mitroff et al,
1996).
2.1 The Early Warning Signs of
Crises
Hidden-ness is an innate characteristic of crises.
Very often crises are hidden underneath a seemingly
normal surface and invisible to us until it already
comes into strong appearance, which is why crises
always take us by surprise. Nonetheless, crises do
not just form overnight. In fact, crises are often
presaged by various visible or invisible signals. As
described by Mitroff et al (1996), “with very few
exceptions, crises send out a trail of early warning
signals before their actual occurrence”. The pity is
that such warning-signs are often widely spread
across the organization and embedded in various
information thus are not so easy to recognize. In
some cases, they are neglected even if they already
present themselves in front of people, or some
already identified warning signs may be
intentionally or unintentionally blocked in its way to
reach the appropriate management attention.
DEFENDING AGAINST BUSINESS CRISES WITH THE HELP OF INTELLIGENT AGENT BASED EARLY
WARNING SOLUTIONS
59

Crisis warning signals may be early or late, visible
or hidden, strong or weak, physical or informational.
In the theory of crisis management, there are
generally four types of crisis early warning pattern:
(i) economic crisis may signal business crisis
(economic life cycle); (ii) trade and industry crisis
may signal business crisis (decline progressively;
excessive production); (iii) business crisis may
signal industry crisis; (iv) catastrophic events may
signal business crisis (Gao and Yuan, 2003). This
very generally informs us where we could look for
signals of crises. However, this is first a too general
a theory to guide further implementation, and second
we regard such signals as only rather “strong” (or
late, some are accidental) instead of weak (or early,
non-accidental) indicators of crises. Obviously it
does not provide enough advice. In fact, in the
literature of crisis management, we often find crisis
cases that are usually a gas leak in chemical plant, a
call from media, or similar signals that are already
pretty strong indicators. Of course it is important to
attend to strong signals; but as important we must
also try to attend to weak signals that are early
indicators. But, after all, where else do early warning
signals come from?
To be able to define what are the possible early
warning signs of certain types of crises that an
organization may especially be vulnerable to is
perhaps the most important step in establishing an
early warning system. In Fink’s words, at the pre-
crisis stage, “small, seemingly insignificant warning
signs can signal an impending disaster for those who
know where to look”. One intuitive solution may be
to look at the “sources of crises” for “sources of
early warning signals”. “Sources of crises” may be
external or internal as mentioned above. Inside a
company, “sources of crises” may be from those
directly related to key aspects in business
“operations” and business “management”. On the
“operation” side, crises may be resulted from all
aspects along the supply and demand chain, for
example: decreased sales, deteriorated financial
status, damaged customer/supplier/partner
relationships, core technology failure linked to key
product and services, etc. On the “management”
side, crises may be resulted from poor decisions and
management actions on key operational factors
along the supply-demand chain or any other aspects
of the corporate life, unexpected harmful
consequences in the implementation of certain
decision management structure, mis-managed public
relations, HR/occupational issues, inappropriate
company culture and corporate values, and so on. In
addition to sources of crises as the direct/immediate
“sources of warning signals”, the close causal-effect
relationships among various early warning factors
need also be attended.
To a crisis management team, the physical “sources
of early warning information” often include internal
and external sources such as: (i) externally: media,
police, health department, government, industry
bodies; (ii) internally: public affairs, health/safety,
security, legal, operations departments (Mitroff et al,
1996). As a complement, early warning information
can also be discovered systematically from large
volumes of sales data, financial data, customer
information, supplier information that can be found
from various credible and complementary sources.
For example, average sale revenue (per employee)
compared with a standard value, its percentage
changes, and sale revenue change compared with
profit change, are often the sales indicators of a
crisis situation. When sales revenue is decreasing, it
may signal some risky situations such as the industry
is in recession; the competitor is growing strong;
customers’ indemnity claim increase; customers
change frequently; customers stop increasing; major
products are not welcome on the market; inventory
increases, etc. Looking at the financial figures,
certain degree of shortage in cash flow can cause
crisis situations (e.g. continuous in deficit for 5 years
together with non-improvement in sales or
continuously decreased sales); increased cost, too
much human resource cost, too much equipment
investment, inappropriate large investments resulted
in to much liability. A crisis can also be the result of
dangerous customers and suppliers. Signals of
dangerous customers include for example absent
managers, key manager goes to the hospital,
customer’s customer turn away, etc. (Gao and Yuan,
2003).
Then, how far we should go to look for early
warning signals? The ideal way is of course try to be
vigilant to all possible leaks of warning signals with
a 360˚ radar monitoring system. This is however
often restricted by the resources available. In Mitroff
et al (1996), they established a concept of “crisis
family”, and suggested that any crisis portfolio of an
organization should contain at least one crisis from
each crisis family. It acknowledges that each type of
crisis may be either the cause or the effect of another
type of crisis. An organization that wants to be well
prepared for dealing with possible major crises must
have the capability to handle at least a crisis
portfolio. This to some extent could be a useful
guidance for limiting the scope of warning signal
searching. Another tool that can be applied to help
us roughly prioritize our attention is the framework
of business life cycles.
ICEIS 2005 - ARTIFICIAL INTELLIGENCE AND DECISION SUPPORT SYSTEMS
60

2.2 Business Life Cycle and Crisis
Early Warning
The theory of business life cycle tells us that, with
the passage of time, in the process of its
development, all businesses will go through various
life-cycle stages from conception to cessation,
similar to the cycle of life of human beings or the
cycle of life for products and services. In 1979,
Ichak Adizes proposed his ten-stage corporate life
cycle model describing business development as
including the stages of courtship, infancy, go-go,
adolescence, prime, stability, aristocracy,
recrimination, bureaucracy, and death. Some other
models describe a business life cycle as having
seven stages including the seed stage, start-up stage,
growth stage, established stage, expansion stage,
decline stage and exit stage (Zahorsky, 2003), or
four phases including Start-Up (covers Courtship,
Infancy), Growth (covers Go-Go, Adolescence,
Prime), Maturity (covers Stable, Expansions), Aging
and Decline (or Transition), then return to the Start-
Up stages. Each stage of the business life cycle may
not occur in chronological order. Some businesses
quickly go from Start-up to Exit. Some may choose
to avoid expansion and stay in the established stage.
External environment factors (such as economic life
cycle, industry life cycle) can directly affect the
course of a business life cycle and cause shifts in life
cycles. Whether a business is a glowing success
depends a lot on its ability to adapt to its changing
life cycles.
Business life cycle is the more stable rule in the
corporate life. The characteristics of a business at
each phase are usually well known. For example, at
the early stage, resources are very limited and cost
or expenses control is critical. The growth stage is
characterized by strong profits as well as emerging
competition. At its established stage, a business has
matured into a thriving company with a place in the
market and with loyal customers. Business life has
become more a routine. But new issues such as
macro economic environment, competitors or
changing customer tastes will call for attention. The
expansion stage is characterized by a new period of
growth into new markets and distribution channels.
But moving into unrelated businesses can be
disastrous. Finally, changes in the economy, society,
or market conditions can cause sustainable decrease
in sales and profits that will lead the business into
the decline stage, which eventually may end a
business (Zahorsky, 2003).
The relevance and importance of business life cycle
to crisis early warning is that the life cycle's
predictable patterns help managers to develop
insight as to what problems need to be corrected
first. Risks or possible crises associated with
different stages of the business life cycle will differ.
At different stages of its life cycle, a business will be
especially vulnerable to different threats. What a
business needs to heed and overcome today will
change in the future. Life stage information can thus
be used as one of the frameworks for prioritizing
signals, or guiding our attention to various signals.
For example, while there is always the possibility of
running into crisis situations, companies over 30
years history and business at the time of transition
can be especially vulnerable and at very high risk
because of reasons such as lack of successor
managers, or outdated rigid systems (in Gao and
Yuan, 2003).
In addition, the life cycle model thus provides useful
basis for understanding more about organizational
change. Especially related with the life cycle
analysis are also the risks arising from the
management of the life cycle: the control of the
speed of business development, especially the speed
of growth and expansion. Crises may happen if a
new product or service moves into growth phase too
slowly, move out of it too quickly, or if a business
cannot move back to growth from the maturity. The
risks associated with mergers and acquisitions are
often high, and at the transition stage, there may
generate very risky situations associated with
business decline, when a crisis may happen if the
business can not exit a particular market in time or it
cannot support a negative cash flow for long enough
(in Gao and Yuan, 2003).
2.3 Early Warning Analysis: from
Warning Signs to Crisis Alerts
Earlier business failure models are mainly financial
models applying statistical methods to financial
ratios as the tool for the analysis of business failure
risks, which would give an early indication of
potential financial difficulties. The most popularly
known conventional models for early warning
include Beaver’s univariate analysis using
dichotomous classification test (1967), and Altman’s
Z-Score model (1968) and its extension, the ZETA
model (Altman, Haldeman and Narayanan, 1977).
The Z-score model is based on a statistical
multivariate discriminant method. The ZETA model
was developed to further improve the forecasting
accuracy of the Z-score model, and to account for
the changes and developments in the macro
economic environment over time. Marked by the
work of Lapedes and Fayber (1987), neural network
DEFENDING AGAINST BUSINESS CRISES WITH THE HELP OF INTELLIGENT AGENT BASED EARLY
WARNING SOLUTIONS
61

models have found wide application in the more
recent studies of business failures, and they have
shown much potential in dealing with non-linear,
complexity, dynamics and uncertainty in forecasting
(in Gao and Yuan, 2003).
Nowadays, it is easy to find numerous studies that
have shown a definite potential of financial ratios as
predictors of corporate financial distress or
bankruptcy, with profitability, liquidity, leverage,
solvency as the most significant indicators. In our
opinion, however, such a financial-figure based
approach is better suited for post-crisis analysis to
gain a better understanding of the forces behind
crises. From crisis early warning point of view, the
financial ratio approach alone would seem too
insufficient because financial figures only reflect the
symptoms or results of problems rather than the real
causes. When the effect of potential problems is
already reflected in financial figures, it may have
become too late to avoid a crisis situation or to have
it quickly under control. There is definitely the need
to watch and analyze as well all other types of
warning signals such as from sales, customers,
suppliers, raw materials, and other critical operating
factors, as described earlier. Besides the primary
early warning indicators, an early warning analysis
should also pay attention to relationships between
various crises, crisis chains, as well as possible
vicious cycles of crises.
3 INTELLIGENT AGENT BASED
EARLY WARNING SOLUTIONS
3.1 The Roles of IT and Intelligent
Agents in Crisis Early Warning
Systems
Early warning, as part of a crisis management and
decision-making process, is a process of problem
identification, rather than alternative design and
choice making. A crisis early warning system is
bound to be a data-driven system. In order to
effectively support decision makers in the early
detection of crisis situations, an early warning
system needs to be first of all an information
management instrument that can constantly acquire,
monitor and analyze the early warning signals of
crises. In addition, the assessment of risks is at the
core of an effective early warning system.
To be able to provide early warnings of emerging
crises in a timely fashion has proven to be very
challenging tasks due to many reasons. First of all,
crises are usually the result of continuous
interactions among multiple risk factors. Due to the
complicated nature of the interactions, there are
inherent difficulties to formally and precisely
capture and describe the causal-effect relationships
between these factors and a crisis situation,
especially when it is acknowledged that such causal-
effect relationships are not always the same but
instead can change over time. This leads to
difficulties in evaluating even the already visible
signals using conventional statistic tools and
mathematical model based on domain theories.
Second, it is not an easy task either to collect
information on early warning signals in the first
place, not to mention to keep them up-to-date all the
time, and to absorb and make sense of the collected
information in time. Modern IT or ICT can
contribute to crisis management by providing
process support, communication support or
intelligence support. For early warning in the age of
information overload, the effective application of
information technology is at the heart of intelligence
analysis. The importance of the role of IT system for
early warning is especially due to the “systematic
nature” of crisis management. A computer based
early warning systems can enforce the early warning
guidelines into a consistent system. It can also
remove some of the blocks between the crisis
manager and the early warning signals. The
collected information related with crisis is a valuable
resource for supporting the assessment of damages
when crises do happen.
Although various efforts have been made to and
much progress has been achieved in improving the
information search, query, retrieving and scanning
techniques, they only helped in coping with the issue
of high irrelevance, particularly at the “file” level
instead of the “information” level. As far as the
content is concerned, as far as the effective
absorption and use of information is concerned,
severe information overload remains a critical issue.
Especially, it is more often the case that the
collected data are in the form of long text or large
document collections that contain much qualitative
information rather than numeric data, or that contain
a mixture of both. It can become an extremely time-
consuming, labor-intensive and costly process (if not
an impossible process) when trying to grasp the
major content of large amount of news and reports,
to extract important information from them, and
subsequently make effective use of them in decision
making.
An intelligent agent is a computational program that
inhabits in a computing environment, and act on
behalf of users to accomplish delegated tasks.
ICEIS 2005 - ARTIFICIAL INTELLIGENCE AND DECISION SUPPORT SYSTEMS
62

Central to the notion of software agents are the
automation of work and the automation of computer
usage. They are able to take initiative, to solve
problem without direct intervention and constant
step-by-step guidance from the user. They are able
to customize assistance and service, according to
what is learned about the user, and are able to
improve performance based on previous experience.
When deemed appropriate, an agent is able to
interact and collaborate with other agents, or humans
in order to complete their own problem solving and
to help others with their activities; agents work in
background and serve round-the-clock. Agent
technology has found various applications in broad
fields. Especially in information and knowledge
management related tasks, agent based solutions
have played many important parts and have much
potential. In comprising a crisis early warning
system, agents that act on behalf of human
surveillance can save human resources. With the
deployment of multiple software agents, we can
make an early warning system move one step closer
to 360˚ comprehensive radar system.
3.2 Agent Based Early Warning
Systems: A Framework
Based on the domain analysis in previous sections,
we propose to adopt a business life-cycle analysis
based approach to early warning. The business life
cycle model provides the information basis for the
correct selection of early warning methods and
models that may take into account different groups
of early warning signals or implements different
analytical methods. Effective early warning requires
the acquisition and delivery of critical information in
a robust fashion. Technically, such instruments
should be able: (1) to quickly sum up key
information from text data sources (e.g. news text or
business/government reports); (2) to quickly extract
specific pieces of information from text documents;
(3) to discover from large structured data bases the
meaningful hidden information or patterns; and (4)
to make these functionalities easily accessible and
usable for decision makers. In the figure below, we
present a conceptual framework for an intelligent
agent based early warning system. The Text
Summarization Agent and Information Extraction
Agent are responsible for the raw data supply for the
early warning system. They will draw up the
relevant data from credible sources (text sources
such as media, business reports; structured sources
such as corporate databases). There can deploy a
number of text summarization agents as well as
information extraction agents as needed. These
agents will embrace the differences in sources inside
themselves, but will communicate with a
coordination agent to integrate the data, or with the
user to present the summary of long text documents.
Business Life Cycle Choose Early Warning Models
Text
collection
Structured
data sources
Information scanning: prepare
information on early warning signals
defined by selected early warning
models
Text Summariztaion
Agent
Information Extraction
Agent
Data for
early warning
analysis
Intelligent Data Analysis
Agent
Early Warning Agent
DEFENDING AGAINST BUSINESS CRISES WITH THE HELP OF INTELLIGENT AGENT BASED EARLY
WARNING SOLUTIONS
63

The Intelligent Data Analysis Agent will conduct
mining operations to derive the value of early
warning indicators, to prepare the input to early
warning agent. The Early-Warning Agent will apply
risk assessment and crisis identification rules (e.g.
fuzzy if-then rules) to generate risk warnings if there
are crises impending, or establish the degree of
exposure or vulnerability to dangerous consequences
even if there is no crisis situation yet. Again multiple
agents can be deployed to specialize in different
types of crises, such as proposed in the crisis
portfolio by Mitroff et al (1996).
3.3 Soft Computing Methods for
Early Warning Analysis
Fuzzy logic and neural network methods have found
many successful applications in risk assessment
from financial markets, environment control, project
control, health care. Fuzzy method allows the use of
qualitative indicators. In the analysis of early
indicators of business crises, we often have very
limited knowledge of the quantitative relationships
between the variables as well as between the
variables and a crisis situation, and these
relationships may be intrinsically non-linear and
very complex. Especially, it can be a very difficult
task to tell when does the sum of various business
difficulties become critical enough to trigger sudden
worsening of business situation, i.e. the occurrence
of crises. We believe that there is a good potential in
applying intelligent data analysis methods such as
fuzzy set theory and SOM (Self Organizing Maps)
based clustering methods to automatically determine
the “critical levels” or threshold levels of early
warning indicators based on empirical data.
Clustering analysis help us to group data points that
represent multidimensional data into natural clusters
such that the clusters will show an underlying
pattern hidden in the data. There is no need to
assume anything about the functional form of a
hypothesized relationship between the variables.
Fuzzy clustering analysis then allows us to describe
classes or clusters that can at best be imprecisely
articulated. Data points can be partitioned into
overlapping natural groups (i.e. fuzzy clusters),
where each data point belongs to each cluster to
certain degree that is specified by a membership
value. With these analyses of early warning
information, we shall also be able to add to existing
theoretical knowledge of financial crises and
business crises and the descriptive type of case
studies around them. In our previous work, we have
applied the FCM fuzzy clustering method to group
multiple economic time series data into overlapping
clusters, and at the same time to inform us the
degree of membership for each data point in each of
the clusters. This not only helps us to grasp the
feature of economic fundamentals over the crisis
period, but also makes it possible for us to tell the
extent to which one or more warning signals were
perhaps visible already during the normal
development period. As frequently used subjective
categories in crisis analysis are also often imprecise
and do not have sharp boundaries, a fuzzy clustering
approach thus seems to provide a simple and natural
way to capture hidden patterns relationships in an
easier and very flexible way. There is no reason why
such a method cannot be applied in analyzing
business crises.
SOM is a powerful neural network technique that is
very good at dealing with vast multidimensional
data, finding structures in them and visualizing the
data on special map displays. It is better suited in
cases when there are sufficient large empirical data
sets. Comparing with other computer based
analytical methods, fuzzy methods excel in the
ability for dealing with uncertainty, imprecision,
qualitative information, meaning and knowledge
representation for language processing. Neural
networks then offer us powerful means to handle
large amount of data units and to handle complexity.
SOM models, with self learning and self
organization capability, can improve its result
according to changes in the environment, offer
solutions in situations when explicit mathematical
models are difficult to be determined.
Finally, to prioritize the identified crises, multi
criteria ranking facility can be used (based on e.g.
likelihood of damage, criticality of the asset, severity
of the threat, and so on).
4 CONCLUSIONS
It is perhaps true that the most effective way of
being well-prepared for crises is to “foster a
corporate culture that people feel free to express
their opinions” (Mitroff et al, 1996). This however
does not imply that the establishment of a
formalized early warning system is less important.
The importance of early warning system cannot be
overstated. Too often people realize that even
stronger warnings of impending crisis go unheeded
in many organizations. In reality, some crises are
inevitable no matter how well prepared an
organization is. This, however, does not imply any
insignificance of early warning and prevention of
crisis either. It only indicates that “complete
ICEIS 2005 - ARTIFICIAL INTELLIGENCE AND DECISION SUPPORT SYSTEMS
64

prevention” of crisis is not the goal. An early
warning system can also help a business in the
process of recovering from crisis as fast as possible
and to learn from past crises.
An effective early warning system can be expensive
and time-consuming to establish. This does not
mean that businesses should abandon the idea of
early warning simply because of that. In the practice
of business management, there is a pressing need for
good information management instruments that can
constantly acquire, monitor and analyze the early
warning signals of business crises and economic
crises, thus effectively support decision makers in
the early detection of crisis situations. The key
questions are what kind of early warning system a
company should have, to watch out for what signals,
for which crises? With the development of advanced
computing methods and information technology,
there bring new opportunities for the construction of
such instruments. In this paper, we proposed the use
of business life cycle model as a larger framework of
guidance for an early warning system of business
crises. We also developed a framework for an
intelligent agent based early warning system, and
discussed the application of soft computing methods
in the intelligent analysis of early warning
information. This will provide a starting point for the
development of intelligent agent based early
warning solutions.
As Mitroff et al (1996) pointed out, crisis
management both as a research field as well as a
corporate function in general is still new and is
neither completely understood nor systematically
explored. In the process of building crisis early
warning systems, there is also the opportunity for us
use the tool to get a better understanding of the early
warning task and process, as well as other crisis
management issues in an organization.
REFERENCES
Adizes, I., 1990. Corporate Lifecycles: How and Why
Corporations Grow and Die and What to Do About It,
Prentice Hall Press.
Adizes, I., 1999. Managing Corporate Lifecycles, Prentice
Hall Press.
Altman, E.I., September 1968. Financial Ratios,
Discriminant Analysis and the Prediction of Corporate
Bankruptcy. Journal of Finance.
Altman, E.I., Avery R., Eisenbeis R., and J. Sinkey, 1981.
Application of Classification Techniques in Business.
Banking & Finance, JAI Press.
Altman, E.I., R. Haldeman, and P. Narayanan, June 1977.
ZETA Analysis: A New Model to Identify Bankruptcy
Risk of Corporations. Journal of Banking and
Finance.
Augustine, N.R., 2000. Managing the Crisis You Tried to
Prevent. in Harvard Business Review on Crisis
Management, Harvard Business School Press.
Beaver W., 1966. Financial Ratios as Predicators of
Failure. Empirical Research in Accounting Selected
Studies. Supplement to Journal of Accounting
Research. pp. 71-111.
Zahorsky D., 2003. Small Business Information,
http://sbinformation.about.com/cs/marketing/a/a04060
3.htm
Gao and Yuan, 2003. Early Warning of Corporate Crises
(in Chinese), China Economic Publishing House
Mitroff I.I., C.M. Pearson and L.K. Harrington, 1996. The
Essential Guide to Managing Corporate Crises,
Oxford University Press.
DEFENDING AGAINST BUSINESS CRISES WITH THE HELP OF INTELLIGENT AGENT BASED EARLY
WARNING SOLUTIONS
65